This chapter examines the challenges that Japan faces in implementing initiatives to maximise the benefits of AI in the workplace, such as training and worker consultation. Japan appears to trail other countries when it comes to the implementation of such initiatives. Japan also needs to improve the inclusiveness of these initiatives. Older workers and non-regular employees in particular are less likely to participate in training to work with AI and are less likely to be consulted when new technologies are introduced in the workplace. Older workers also have lower levels of trust in their companies using only safe and trustworthy AI. Better engaging older workers in these initiatives is particularly important in the context of population ageing. Under the new AI legislation, Japan should continue to advance AI-related labour market policies with a dual focus: promoting the use of AI while simultaneously addressing the risks associated with its application.
Artificial Intelligence and the Labour Market in Japan
Abstract
This chapter identifies the challenges that Japan faces in maximising the benefits of using AI in the workplace, focusing on four key initiatives: training and learning, worker consultation, guidelines for the use of GEAI at work, and ensuring the safety and trustworthiness of AI technologies in the workplace. The findings indicate that Japan needs to promote these initiatives further and broaden their inclusiveness to reach a wider range of workers.
Company-led AI training and employee self-learning to work with AI.
With 30% of AI users reporting receipt of company training to help them work with AI, Japan ranks the lowest among the eight countries surveyed (the highest being Ireland, with 69%). The greatest gaps between Japan and the other countries are for: male, middle‑aged and older AI users. In addition, the proportion of AI users in Japan reporting participation in company training is only 21.2% among those in non-regular employment, compared to 34.1% among those in regular employment. Japanese companies must expand training and financial support to maximise AI’s benefits in the workplace, while also paying attention to improving inclusiveness. The MHLW already provides subsidies for AI-related company training through the Human Resources Development Subsidies (HRDS), but enhancing the accessibility of these subsidies could encourage more companies to implement such training programmes.
The Educational Training Benefits (ETB) has played an important role in supporting a wide range of workers who pursued reskilling or upskilling to work with AI in Japan. However, among AI users who undertook reskilling or upskilling, the only 55% used ETB, indicating substantial room for further uptake. MHLW should raise awareness of the scheme, thereby removing barriers to participation – especially for older and non-regular workers – and enhancing inclusiveness. Furthermore, to strengthen ETB support for AI-related reskilling and upskilling, it is important to broaden the pool of eligible training providers while ensuring, through inter-ministerial co‑ordination, the availability of high-quality, appealing programmes that meet the diverse needs of learners.
Econometric analysis reveals regional disparities in the proportion of Japanese AI users who report having access to resources for learning to work with AI. If companies are considering providing financial assistance to support employees’ studies at universities or other educational and training institutions or are outsourcing off-the‑job training (Off-JT) related to AI, relevant resources within the region need to be made available. The Regional Consortiums for Vocational Abilities Development Promotion (RCVADP), established in each prefecture, could serve as effective platforms through which stakeholders and relevant organisations can collaborate to expand the range of Off-JT related to AI, and the financial support options that companies can provide to their employees to work effectively with AI, in alignment with the needs of local businesses. Furthermore, with regard to the ETB, efforts should continue to expand the number of online courses – while ensuring the quality and effectiveness of training – in order to prevent regional disparities.
Econometric analysis reveals that Japanese AI users working at companies experiencing labour shortages are less likely to report having access to resources for learning to work with AI. When companies experience labour shortages, it may become more difficult not only for them to allocate personnel for organising company training or to provide financial support for training, but also for employees themselves to engage in self-directed reskilling or upskilling due to increased workloads. Therefore, the MHLW needs to continue supporting companies in securing human resources by promoting the Human Resource Development Subsidies, strengthening job-matching services at PES (known as Hello Work in Japan), and raising awareness of good practices in employment management that help retain employees.
Worker consultation on the use of new technologies in the workplace.
With 42% of AI users in Japan saying their employers consult them on the use of new technologies, Japan ranks lowest among the eight countries surveyed. The highest share of AI users consulted can be found in Germany (70%). Since worker consultation is associated with better outcomes for workers, it will be important to promote it in Japan. In addition, from an inclusiveness perspective, it will be important that such consultation is extended to middle‑aged and older workers, as well as to those in non-regular employment. To foster workplace cultures and practices that promote communication between employers and employees when introducing new technologies, it is important for MHLW to raise awareness – using concrete examples and evidence – of the value of such initiatives, which have the potential to maximise the positive impact of AI on job performance and working conditions and deliver mutually beneficial outcomes for both workers and employers.
Internal rules or guidelines to ensure employees can use GEAI appropriately in their work.
Among Japanese GEAI users, only 34.8% report that internal rules or guidelines to support appropriate use of AI at work have been established, and the evidence suggests Japanese companies may be lagging behind internationally in this area. Among GEAI users whose companies have established such guidelines, only 34.1% indicate that they understand their contents almost perfectly. Furthermore, among GEAI users who report at least a partial understanding of the guidelines, only 37.5% say there were aspects they were not fully complying with. The Japan Deep Learning Association (JDLA) has developed model guidelines for the use of GEAI, freely available to all on its website. Japanese companies should actively use these resources to promote the development of guidelines tailored to their operational contexts. Moreover, it is increasingly important for companies to ensure that employees fully understand and comply with them.
Ensuring the safety and trustworthiness of AI technologies introduced in the workplace.
With 66% of AI users saying they trust their employers to use only safe and trustworthy AI, Japan ranks lowest among the eight countries surveyed. The highest share of AI users expressing such trust can be found in the United States (86%). Among workers who do not use AI, Japan also ranks lowest regarding the proportion reporting such trust, assuming AI were to be introduced in their workplace. AI non-users in Japan are 27.4 p.p. more likely to report a lack of trust than AI users. This tendency not to trust their employer is more pronounced among non-users who are male, middle‑aged and older workers, those employed in SMEs or companies experiencing labour shortages, and those in lower-skilled occupations.
The MHLW should continue to promote AI-related labour market policies with a dual focus: encouraging the use of AI while simultaneously addressing the associated risks. Specifically, MHLW should compile good practices from companies and disseminate this information – together with relevant data and evidence – in a manner that is easily accessible to a wide audience. Subsequently, policymakers could consider developing guidelines to support both employers and workers in addressing the benefits and risks associated with workplace AI use.
This chapter explores the challenges associated with appropriately preparing for the future of work in Japan, with a particular focus on following four key initiatives aimed at maximising the benefits of AI integration into the workplace for improving job performance and working conditions.
1. Providing company training and financial support to help employees work effectively with AI, as well as encouraging employees self-initiated learning.
2. Promoting worker consultation on the use of new technologies in the workplace.
3. Establishing internal rules or guidelines to ensure employees can use GEAI appropriately in their work and maintaining ongoing communication following its implementation.
4. Actively building employee trust – through various forms of communication and the use of external third-party risk assessment bodies – that their company will introduce only safe and trustworthy AI technologies in the workplace.
4.1.1. Challenges related to company-provided training and financial support
A comparison of company training and financial support provided to help employees work effectively with AI reveals that such support is less commonly offered to AI users in Japan than in the other surveyed countries (Figure 4.1). Specifically, in the finance and insurance sector, only 26.9% of Japanese AI users report receiving company-provided training, compared to an average of 57.5% across the seven other OECD countries surveyed. In the manufacturing sector, the figures are 33.3% for Japan and 53.3% for the seven other OECD countries surveyed. A country-level comparison further shows that, in both sectors, Ireland has the highest proportion of AI users reporting the receipt of company-provided training (69%), while Japan ranks lowest among the eight countries surveyed, with the proportion hovering around 30% (Annex Figure 4.A.1).
Other surveys also suggest that Japan may be lagging behind in the provision of AI-related training (Box 4.1). The relatively modest improvements in job quality due to AI observed in Japan (Figure 2.1), may therefore explained in part be by the slower implementation – compared to other surveyed countries – of company-provided training and financial support to help employees work effectively with AI. These findings suggest that Japanese companies need to further promote the provision of such training and financial support. In addition, while the MHLW currently provides subsidies for various types of employer-led training – including AI-related training – through the Human Resources Development Subsidies (HRDS), it is also important to explore ways to make these subsidies more accessible for companies considering the provision of training to help employees work collaboratively with AI.
Percentage of AI users
Note: AI users were asked: “Has your company provided or funded training so that you can work with AI? (Yes; No; Don’t know)”
Source: OECD worker survey on the impact of AI on the workplace (2022), JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
Gender and company-provided training
The gap between Japan and the average of the seven other countries surveyed in the proportion of AI users reporting access to company training and financial support to help employees work effectively with AI is notably larger among male AI users than among female AI users (Figure 4.3). In the finance and insurance sector, the proportion of female AI users in Japan reporting receipt of company training or financial support is 22.4 p.p. lower than the average across the seven other countries surveyed, whereas the corresponding figure for male AI users is 36.5 p.p. lower. Similarly, in the manufacturing sector, the proportion of female AI users in Japan reporting receipt of such support is 14.7 p.p. lower than the seven‑country average, whereas the corresponding figure for male AI users is 21.7 p.p. lower. These findings indicate that, while Japan is characterised by a relatively smaller gender gap in the provision of company training and financial support to help employees work effectively with AI, there is a need to further strengthen efforts targeting male AI users in order to catch up with other countries, while ensuring that the gender gap does not widen.
According to an OECD study on the use of generative AI in SMEs (2025[1]), the share reporting that their employees participate in AI-related training is 11.3% in Japan – the lowest among the countries surveyed (Figure 4.2). This report suggests several possible reasons why AI-related training is not yet common, even among SMEs using generative AI. One is that SMEs may not perceive a need to offer training, since generative AI can be used through conversational interfaces without requiring coding knowledge. Another possibility is that generative AI has advanced so rapidly that SMEs have not yet had the opportunity to organise training. In particular, the report suggests that the low participation rate in AI-related training in Japan may be partly due to the fact that many Japanese SMEs are experiencing a lack of skills or experience among their staff, as well as labour shortages. Across OECD countries, time constraints due to work are the most cited barrier to participation in job-related non-formal learning, with the share of adults citing this reason in Japan exceeding the OECD average.
In general, compared to larger companies, SMEs face higher unit costs per worker and may lack the resources to finance training opportunities. Moreover, with fewer employees, SMEs have less flexibility to release staff from revenue‑generating activities to undertake training. Concerns around worker poaching, where workers leave for better opportunities after receiving training, further discourage investment in training. SMEs typically offer lower wages, less attractive working conditions, and fewer career advancement opportunities than larger companies. As a result, trained employees may be more inclined to seek employment elsewhere, reducing the returns to training investments for small businesses.
Percentage of AI users, by country
Note: AI users were asked: “Has your company provided or funded training so that you can work with AI? (Yes; No; Don’t know)”
Source: OECD worker survey on the impact of AI on the workplace (2022), JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
Percentage of AI users, by country, by gender
Note: AI users were asked: “Has your company provided or funded training so that you can work with AI? (Yes; No; Don’t know)”
Source: OECD worker survey on the impact of AI on the workplace (2022), JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
Age group and company-provided training
The gap between Japan and the average of the seven other countries surveyed in the proportion of AI users reporting access to company training and financial support is notably larger among middle‑aged and older AI users than among younger AI users (Figure 4.4). In the finance and insurance sector and the manufacturing sector, the proportion of younger AI users (aged 15‑34) in Japan reporting receipt of company training or financial support is 18.2 p.p. lower than the average across the seven other countries surveyed, whereas the corresponding figure for middle‑aged AI users (35‑49) is 24.2 p.p. lower, and for older AI users (50 and above) is 29.7 p.p. lower. In addition, while the average across the seven other countries surveyed shows little variation by age group, Japan exhibits comparatively large age‑related disparities in access to company-provided training and financial support. These findings indicate that, in order for Japan to catch up with other countries in the provision of company training and financial support to help employees work effectively with AI, it is necessary to further strengthen efforts targeting middle‑aged and older AI users.
Percentage of AI users, by country, by age group
Note: AI users were asked: “Has your company provided or funded training so that you can work with AI? (Yes; No; Don’t know)”
Source: OECD worker survey on the impact of AI on the workplace (2022), JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
Among other worker characteristics, there may be challenges in access to company-provided training and financial support for working with AI, particularly for AI users in non-regular employment. Among AI users, 34.1% of those in regular employment report that their company provides such training, whereas the figure drops significantly to just 21.2% for AI users in non-regular employment. These figures indicate that Japanese companies should enhance their efforts to support non-regular workers in order to improve the inclusiveness of company training and financial assistance aimed at helping employees work effectively with AI.
4.1.2. Challenges concerning the support for workers’ voluntary reskilling and upskilling
In Japan, the MHLW supports the cost of training provided by employers through the Human Resources Development Subsidies (HRDS), while self-initiated reskilling and upskilling by workers is supported via the Educational Training Benefits (ETB). Under this scheme, a portion of the training costs is reimbursed to workers who successfully complete courses designated by the Minister of Health, Labour and Welfare. The designated courses are categorised into three types based on their level and purpose, each with a different subsidy rate. More specifically, (1) the Vocational Education and Training Benefits category provides the highest level of support, reimbursing up to 80% of the training cost (with an annual cap of 640 000 JPY).1 (2) the Specifically Designated Education and Training Benefits category covers up to 50% of training costs (with an annual cap of 250 000 JPY),2 while (3) the General Education and Training Benefits category reimburses up to 20% (with an annual cap of 100 000 JPY). As of 1 April 2025, the number of designated courses available stood at 3 220 for category (1), 1 016 for category (2), and 2 664 for category (3). In particular, courses that include the term “AI” in their title are predominantly found within the Specified Practical Education and Training category. Additionally, while some courses listed on METI’s Manabi DX platform are not eligible for support under the ETB, the portal serves as a curated platform that provides a wide range of learning content related to digital skills, including free courses.
This report aims to examine the extent to which Japanese AI users make use of the ETB when undertaking reskilling or upskilling, in order to identify key challenges. As a starting point, it organises the status of reskilling or upskilling efforts by distinguishing between AI users, AI non-users (AI adopters), and AI non-adopters. An analysis based on this classification reveals that AI users in Japan are more likely to have engaged in reskilling or upskilling to work with AI, compared to AI non-users (AI adopters) and AI non-adopters (Figure 4.5). Specifically, 34.7% of Japanese AI users report undertaking reskilling or upskilling in 2023 to work with AI, while 23.5% report engaging in other forms of skill development. Among AI non-users (AI adopters), 10.6% report undertaking reskilling or upskilling for AI-related purposes, and 22.6% report engaging in other types of training. Furthermore, 4.0% of non-AI adopters report undertaking reskilling or upskilling to work with AI, while 19.8% report engaging in other forms of training.
Percentage of all employees
Note: All employees were asked: “In the last year, did you engage in reskilling/upskilling? (Yes; No)” Employees who answered that they engaged in reskilling/upskilling in 2023 were asked: “In the last year, did you engage in reskilling/upskilling to work with AI? (Yes; No)”
Source: JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
Utilisation of Educational Training Benefits by AI usage status
Japanese workers who undertook reskilling or upskilling to work with AI are more likely to report that they used ETB than those who undertook reskilling or upskilling for other purposes (Figure 4.6). Specifically, among those who engaged in reskilling or upskilling to work with AI, the proportion of workers who made use of ETB is 55.4% for AI users, 52.9% for AI non-users (AI adopters), and 52.1% for AI non-adopters. Conversely, among those who engaged in reskilling or upskilling for purposes other than working with AI, the proportion of workers who made use of ETB is 23.2% for AI users, 15.8% for AI non-users (AI adopters), and 7.9% for AI non-adopters. These figures suggest that ETB plays a relatively important role in supporting workers engaging in reskilling or upskilling to work with AI. However, the average utilisation rate of ETB among those reskilling or upskilling to work with AI stands at 53.5%, which indicates that there remains considerable room for improvement in policy uptake. To further enhance support through the ETB for individuals undertaking reskilling or upskilling to work with AI, it will be essential to expand and diversify the pool of eligible training providers, including companies and universities. In this context, the MHLW policymakers must work in close co‑ordination with other relevant ministries to ensure the availability of high-quality and appealing training programmes. Moreover, establishing a virtuous cycle – where an increasing number of motivated learners encourages greater participation from reputable training providers – will be vital for maximising both the reach and the impact of the scheme.
Percentage of employees who report that they engaged in reskilling/upskilling in 2023
Note: Employees who answered that they engaged in reskilling/upskilling in 2023 were asked: “In the last year, did you use the 'Educational Training Benefits' when you reskilled/upskilled?(Yes; No) ” The figure shows the proportion of employees who responded “Yes”. The “Educational Training Benefits” system is designed to support the independent skill development and career formation of working people by subsidizing part of the course fees when they complete education and training specified by the Minister of Ministry of health labour and welfare.
Source: JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
Utilisation of Educational Training Benefits by worker characteristics
There are disparities in access to the ETB depending on the characteristics of workers who engage in reskilling or upskilling to work with AI, with particularly notable gaps observed among older workers and those in non-regular employment (Figure 4.7). Specifically, the utilisation rate of ETB among younger workers is 57.6%, whereas the corresponding rate is 53.2% for middle‑aged workers and 46.7% for older workers. Similarly, the utilisation rate of ETB among regular workers is around 55.6%, whereas the corresponding rate for non-regular workers is around 42.5%. In contrast, there are no substantial differences in the utilisation rate of ETB for reskilling or upskilling to work with AI by gender, educational background, or income level in the 2023. Notably, relatively higher utilisation rates were observed among workers with disabilities and those balancing work with childcare or long-term care responsibilities. These findings suggest that, as MHLW policymakers seek to strengthen support through the ETB for reskilling and upskilling aimed at working with AI, it is essential to simultaneously enhance the inclusiveness of the scheme – particularly for older workers and those in non-regular employment – by further raising awareness and removing barriers to participation.
Percentage of employees engaged in reskilling/upskilling to work with AI, by gender, age, employment status, education, annual income, disability/caregiving status
Note: Employees who answered that they engaged in reskilling/upskilling in 2023 were asked: “In the last year, did you use the 'Educational Training Benefits' when you reskilled/upskilled? (Yes; No)” The figure shows the proportion of employees who responded “Yes”. The “Educational Training Benefits” system is designed to support the independent skill development and career formation of working people by subsidizing part of the course fees when they complete educational training specified by the Minister of Ministry of health labour and welfare. The figure for “University degree” is the total of four‑year university and graduate school. The annual income figures for 2023 are before taxes and social security contributions have been deducted. “Low” is classified as Below JPY 2 000 000. “Middle” is classified as over JPY 2 000 000 and below JPY 8 000 000. “High” is classified as over JPY 8 000 000.
Source: JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
Access to AI learning resources by region
The proportion of AI users reporting access to resources for learning how to work with AI – such as company-provided training and financial support, AI-related courses covered by the ETB, and free training opportunities – varies by region (Figure 4.8). Specifically, it is highest in the Hokkaido/Tohoku region at 53.6%, and lowest in the Hokuriku/Tokai region at 39.3%, representing a regional gap of 14.3 p.p. In addition, the results of a generalised ordered logit model that controls for several individual attributes (gender, age group, educational, employment status, company size, labour shortages or excesses, and occupation) reinforce the possibility of regional disparities in Japanese AI users’ responses to the question of whether they have access to resources to learn to work with AI, based on a five‑point scale3 ranging from “Strongly disagree” to “Strongly agree” (Annex Table 4.A.1). Using AI users in the Hokkaido/Tohoku as the reference group, those in the Kinki are 5.4 p.p. more likely to respond “Somewhat disagree” and 5.8 p.p. less likely to respond “Strongly agree” to this question. Similarly, those living in the Chugoku/Shikoku region are 6.6 p.p. more likely to respond “Strongly disagree” and 10.1 p.p. less likely to respond “Strongly agree”. While no statistically significant differences were observed for users in South Kanto and Kyushu/Okinawa, those in North Kanto and Hokuriku/Tokai also showed a tendency to be less likely to agree that such resources are available.
To narrow regional disparities, it is essential for MHLW policymakers to encourage local companies to actively consider making use of the Human Resources Development Subsidies (HRDS), particularly where financial constraints pose a barrier to providing training for working with AI. Moreover, if companies are considering providing financial assistance for employees to pursue study at universities or other educational and training institutions, or commissioning external Off-JT (Off-the‑job training) related to working with AI, it is necessary to strengthen the availability of relevant resources within the region. At present, each prefecture is required to organise a Regional Consortiums for Vocational Abilities Development Promotion (RCVADP), which brings together local stakeholders and institutions4 to share useful information on vocational skills and to design and implement public vocational training that meets local needs and conditions. These councils have the potential to serve as effective platforms through which stakeholders and relevant organisations can collaborate to expand the range of Off-JT related to working with AI, as well as the financial support options that companies can provide to their employees to work effectively with AI, in alignment with the needs of local businesses. Furthermore, employer organisations participating in these councils should play a key role in clearly and accessibly disseminating information about the availability of such resources to companies in the region.
As noted above, the ETB often designates reskilling or upskilling programmes related to working with AI as Vocational Education and Training Courses. These courses are increasingly being offered online using e‑learning methods, which helps mitigate regional disparities in access. However, some of these courses still require in-person attendance. As the adoption of AI in the workplace progresses, the demand for self-initiated reskilling or upskilling among workers is expected to increase further. To address this rising demand while reducing regional disparities, the MHLW policymakers, in co‑operation with relevant ministries and agencies, should continue to expand the number of online courses, while ensuring the quality and effectiveness of training. At the same time, efforts should be sustained to increase the availability of in-person courses across various regions, so that such opportunities are not overly concentrated in specific areas. As a supplementary point, it is also worth noting that the Basic Policy on Economic and Fiscal Management and Reform 2025 underscores the Japanese Government’s commitment to expanding the range of digital skills courses – including those related to AI – that are eligible for Educational Training Benefits.
Percentage of AI users, by residential area
Note: AI users were asked: “Please answer your perception of the impacts of AI in relation to the skills required in your occupation; I have the resources to learn to work with AI. (Strongly agree; Somewhat agree; Neither agree nor disagree; Somewhat disagree; Strongly disagree; I don't know)” The figure shows the proportion of AI users who (strongly or somewhat) agreed that they have the resources to learn to work with AI.
Source: JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
Access to AI learning resources and labour shortages
When companies experience labour shortages, it may become more difficult not only for them to allocate personnel for organising company training or to provide financial support for training, but also for employees themselves to engage in self-directed reskilling or upskilling due to increased workloads (Figure 4.9). 52.8% of AI users working in companies with an appropriate staffing level or labour excess agree with the statement “I have the resources to learn to work with AI,” compared to only 46.3% of those whose companies are experiencing labour shortages. In addition, the results of a generalised ordered logit model that controls for several individual attributes (gender, age group, educational background, employment status, company size, residential area, and occupation) reinforce the possibility that AI users working in companies facing labour shortages are less likely to report having access to resources to learn to work with AI, based on a five‑point scale ranging from “Strongly disagree” to “Strongly agree” (Annex Table 4.A.2). Using AI users working at companies with appropriate staffing as the reference group, those working at companies experiencing labour shortages are 3.5 p.p. more likely to respond “Somewhat disagree” and 4.5 p.p. less likely to respond “Somewhat agree” to this question. Therefore, the MHLW needs to continue supporting companies in securing human resources by promoting the Human Resources Development Subsidies,5 strengthening job-matching services at PES (known as Hello Work in Japan), and raising awareness of good practices in employment management that help retain employees.
Percentage of AI users, by manpower status in company, by company size
Note: AI users were asked: “Please answer your perception of the impacts of AI in relation to the skills required in your occupation; I have the resources to learn to work with AI. (Strongly agree; Somewhat agree; Neither agree nor disagree; Somewhat disagree; Strongly disagree; I don't know)” The figure shows the proportion of AI users who (strongly or somewhat) agreed that they have the resources to learn to work with AI.
Source: JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
The proportion of all employees and AI users in Japan who are consulted by their employers when new technologies are introduced in the workplace may be significantly lower than the average observed across the seven surveyed OECD countries (Figure 4.10). Specifically, the proportion of all employees who are consulted is 16.3% in the finance and insurance sector in Japan, compared to 49.1% across the seven surveyed OECD countries. In the manufacturing sector, the corresponding figure is 16.4% in Japan, while the OECD average is 42.4%. Similar differences are observed among AI users. The data also indicate that the largest proportion of OECD employees consulted when new technologies are introduced in the workplace is found in Germany, followed by Ireland, while Japan ranks lowest among the eight countries surveyed (Annex Figure 4.A.2). These findings underscore the need for MHLW policymakers and Japanese companies to foster a workplace culture in which worker consultation is conducted when introducing new technologies, in order to maximise the positive impact of AI.
Percentage of All employees and AI users
Note: All employees were asked: “In your experience, does your employer consult workers or worker representatives regarding the use of new technologies in the workplace?"
Source: OECD worker survey on the impact of AI on the workplace (2022), JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
4.2.1. Gender and worker consultation
The gap between Japan and the average of the seven other countries surveyed in the proportion of AI users reporting that they are consulted by their employers when new technologies are introduced tends to be slightly larger among male AI users than among female AI users (Figure 4.11). Specifically, in the finance and insurance sector, the proportion of female AI users in Japan reporting having been consulted is around 29.2 percentage points lower than the average across the seven other countries surveyed, whereas the corresponding figure for male AI users is around 32.1 percentage points lower. Similarly, in the manufacturing sector, the proportion of female AI users in Japan reporting having been consulted is around 5.7 percentage points lower than the seven-country average, whereas the corresponding figure for male AI users is around 10.3 percentage points lower. These findings indicate that, while Japan is characterised by a relatively smaller gender gap in the implementation of worker consultation when introducing new technologies, there is a need to further strengthen efforts targeting male AI users in order to catch up with other countries, while ensuring that the gender gap does not widen.
% of AI users, by country, by gender
Note: Al users were asked: “In your experience, does your employer consult workers or worker representatives regarding the use of new technologies in the workplace?"
Source: OECD worker survey on the impact of AI on the workplace (2022), JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
4.2.2. Age group and worker consultation
The gap between Japan and the average of the seven other countries surveyed in the proportion of AI users reporting that they are consulted by their employers when new technologies are introduced in the workplace tends to be larger among middle-aged AI users than among younger or older AI users (Figure 4.13). Specifically, in the finance and insurance sector and the manufacturing sector, the proportion of younger AI users (aged 16–34/15–34) in Japan reporting having been consulted is 13.5 percentage points lower than the average, and for older AI users (50 and above) is 12.7 percentage points lower than the average across the seven other countries surveyed , whereas the corresponding figure for middle-aged AI users (35–49) is 19.6 percentage points lower. These findings indicate that, in order for Japan to catch up with other countries in terms of consulting workers when introducing new technologies in the workplace, it is necessary to further strengthen efforts targeting middle-aged.
% of AI users, by country, by education
Note: AI users were asked: “Has your company provided or funded training so that you can work with AI? (Yes; No; Don’t know)”
Source: OECD worker survey on the impact of AI on the workplace (2022), JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
4.2.3. Worker consultation by worker characteristics
There are disparities in access to worker consultation depending on the characteristics of AI users, with particularly notable gaps observed among middle‑aged and older workers, as well as those in non-regular employment (Figure 4.13). Specifically, the proportion of younger AI users reporting that they are consulted stands at 50.6%, whereas the corresponding proportion is 40.6% for middle‑aged users and 38.9% for older users. In addition, the proportion of regular AI users reporting that they are consulted stands at 45.5%, whereas the corresponding proportion for non-regular AI users is 36.5%. In contrast, there are no substantial differences in the incidence of worker consultation during the introduction of new technologies by gender, educational background, or income level as of 2023. Notably, relatively higher consultation rates were observed among workers with disabilities and those balancing work with childcare or long-term care responsibilities. These findings suggest that, as MHLW policymakers and Japanese companies seek to foster a workplace culture in which worker consultation is routinely conducted during the introduction of new technologies, it is essential to simultaneously enhance the inclusiveness of such practices by raising awareness of the importance of including middle‑aged and older workers, as well as those in non-regular employment.
Percentage of AI users, by gender, age, employment status, education, annual income, disability/caregiving status
Note: AI users were asked: “Has your company provided or funded training so that you can work with AI? (Yes; No; Don’t know)” The figure shows the proportion of employees who responded “Yes”. The figure for “University degree” is the total of four‑year university and graduate school. The annual income figures for 2023 are before taxes and social security contributions have been deducted. “Low” is classified as Below JPY 2 000 000. “Middle” is classified as over JPY 2 000 000 and below JPY 8 000 000. “High” is classified as over JPY 8 000 000.
Source: JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
4.2.4. Reasons for lack of labour-management communication
This section considers why employers in Japan often do not engage in communication with employees – such as negotiating or consulting with them, providing them with information, or collecting their views – when introducing new technologies in the workplace. (JILPT, 2023[2]) conducted a survey on the introduction of digital technologies such as AI and labour-management communication. Questionnaires were sent to 10 000 establishments with 30 or more employees across Japan, and 1 925 responses were collected. Among establishments that did not engage in communication with employees when introducing new technologies in the workplace (N=733, 38.1%), a multiple‑response question was used to identify the reasons for this. The most frequently cited reason was “Because it wasn’t a particularly significant decision” (27.4%), followed by “Because it was a management decision, negotiation was not considered necessary” (25.2%), and “Because it was the policy set by the head office and parent company” (20.3%) (Figure 4.14).
An examination of the top reasons reveals that, while some establishments lacked autonomy in decision making due to the prioritisation of decisions made by their head office or parent company, others indicated that employers did not perceive the introduction of new technologies as a particularly important decision, or regarded it as a matter falling under their sole authority. A further reason cited was the absence of any established practice of engaging in communication with their employees. It is important to note that the reason “because it was not a particularly significant decision” may suggest that employers, depending on the specific nature of the new technologies – such as AI – being introduced in the workplace, might be inclined to engage in communication with employees if they judge that the introduction could have a significant impact on them. However, the perception that the introduction of new technologies in the workplace is solely a matter of managerial discretion needs to be reconsidered.
As discussed in Chapter 2, worker consultation has the potential to maximise the positive effects of AI on job performance and working conditions. Accordingly, the implementation of such practices is expected to generate mutually beneficial outcomes for both employers and workers. To fully realise these benefits, it is essential that MHLW policymakers actively raise awareness among Japanese employers and workers about the value of these initiatives. This should be done by promoting understanding through concrete examples and evidence‑based explanations, thereby encouraging broader recognition of their importance.
Percentage of establishments that didn’t explain or consult with their employees when introducing new technology in the workplace
Note: Establishments that didn’t explain or consult with their employees when introducing new technology in the workplace: “What were the reasons for not explaining or consulting with employees?” Respondents could select multiple answers. The results show the situation as of 1 January 2023.
Source: JILPT survey on the introduction of digital technologies such as AI and labour-management communication (2023).
In Japan, various organisations have begun to develop and publish their own guidelines on the use of GEAI. The Japan Deep Learning Association (JDLA), a general incorporated association, has created a model set of generative AI usage guidelines to help organisations adopt such technologies more smoothly. These model guidelines have been made publicly available on the JDLA website since 2023, free of charge and accessible to anyone. This initiative supports companies and organisations in developing internal rules or guidelines that are tailored to their specific operations. On the other hand, some Japanese companies have established internal rules or guidelines specifically aimed at prohibiting the use of GEAI in the workplace. While the use of GEAI at work has the potential to enhance job quality for employees, there are also concerns about a range of risks, including the security of customer data, infringement of intellectual property rights, and the spread of misinformation. For these reasons, some organisations have chosen to ban its use entirely. This section aims to clarify the current status of such internal rules and guidelines in Japan and to explore whether any challenges or issues arise in relation to their development and implementation.
The proportion of GEAI users in Japan reporting that internal rules or guidelines to support working appropriately with AI have been established is 34.8% among all GEAI users, rising to around 47.2% when including those who say such guidelines are currently under preparation. Among employees using GEAI based on company instruction, the figure is 41.9% (or 51.2% including those who say such guidelines are under preparation), whereas among those using GEAI on their own initiative, it is only 20.0% (or 38.9% including those who say such guidelines are under preparation) (Figure 4.15). These findings indicate that even when employees use GEAI based on company instruction, internal rules or guidelines to support employees in working appropriately with AI are not necessarily well established across many Japanese companies. Furthermore, it is also suggested that Japan’s initiatives may be lagging behind those of other countries (Box 4.2). As confirmed in Chapter 2, the establishment of such guidelines has the potential to maximise the positive impact of GEAI on job performance and working conditions. Therefore, to avoid falling significantly behind other countries, it is important for Japanese companies to promote the development of guidelines tailored to their specific operations, and to maintain ongoing communication regarding the use of GEAI even after its adoption.
Only 1.9% of Japanese employees report that the use of GEAI has been prohibited by their workplace, rising to 7.0% when including those who say it will be prohibited. Among employees using GEAI at the instruction of their company, 43.0% say that internal rules or guidelines prohibiting the use of GEAI are currently being prepared. This suggests that a significant proportion of employees may be facing a shift towards restrictions on the use of GEAI in the workplace. Moreover, among employees using GEAI on their own initiative, 12.8% report that internal rules or guidelines prohibiting the use of GEAI in the workplace have already been established. This finding suggests that, despite their workplace banning GEAI, a certain proportion of employees may still be using GEAI covertly in the course of their work. Therefore, it is not sufficient for Japanese companies to merely establish internal rules or guidelines prohibiting the use of GEAI for work purposes; a key challenge lies in ensuring that employees fully comply with these rules through effective communication and the thorough dissemination of information.
Percentage of GEAI users and all employees, by whether employees use GEAI under company’s instruction or on their initiatives
Note: All employees were asked: “Does the use of generative AI in the work of employees is prohibited by your company's rules or guidelines? (Yes; No, but my company is now preparing; No, and my company is not preparing; I don’t know)”, GEAI users were asked: “Have internal rules or guidelines been created to ensure that employees use generative AI appropriately in their work? (Yes; No, but my company is now preparing; No, and my company is not preparing; I don’t know)” “Is the use of generative AI in your work company-directed? (business use of generated AI based on company instructions; voluntary business use of generated AI not based on company instructions)”
Source: JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
A study conducted by the OECD (2025[1]) points out that guidelines can support employers in managing the use of generative AI, while also helping workers to use it in a trustworthy manner aligned with company objectives and values, and to avoid potential misuse. According to the report, among SMEs (up to 249 employees) using generative AI, 28.6% have introduced guidelines for employees, whereas the figure is only 8.6% among non-user SMEs. The report also highlights significant cross-country variation in the prevalence of such guidelines: 45.4% of SMEs using generative AI in Germany have established guidelines – the highest among the surveyed countries – while Japan has the lowest share at just 5.3%. The limited provision of internal guidelines may indicate a lack of perceived necessity or awareness among companies. It could also reflect that the use of generative AI tools may be driven by individual initiative rather than top-down strategies.
Although differences in survey subjects and methodologies prevent direct comparison of levels of proportions, the JILPT AI survey indicates that 30.1% of GEAI users in Japan working at companies with up to 250 employees reported that internal rules or guidelines to support employees in working appropriately with AI had already been established. The proportion tends to increase with company size: 35.9% for companies with 250‑999 employees, 37.0% for those with 1 000‑9 999 employees, and 40.1% for companies with 10 000 or more employees. However, there is no strong evidence of a drastic increase in the proportion among companies with over 250 employees. Therefore, even if data from companies with 250 or more employees could be included in Figure 4.16, Japan’s overall figure would likely increase, but its position in international comparisons might not change significantly.
Percentage of SMEs that report having guidelines in place
Note: Respondents were asked: “Does your company have guidelines for staff concerning the use of generative AI?” Results are limited to SMEs using generative AI.
Source: OECD (2025[1]), Generative AI and the SME Workforce: New Survey Evidence, https://doi.org/10.1787/2d08b99d-en.
4.3.1. Understanding and compliance to guidelines
Japanese GEAI users still have considerable room for improvement in their understanding of and compliance with internal rules or guidelines that support employees in working appropriately with AI (Figure 4.17). Among GEAI users for whom such guidelines have already been established, 34.1% report that they understand such guidelines almost perfectly. This proportion rises to 38.1% among those using GEAI based on company instruction, while among those using GEAI on their own initiative, the figure is 23.5%. On the other hand, although only 2.9% of GEAI users for whom these guidelines have already been established report that they have not yet read them, the combined proportion of those who report that they hardly understand them and those who do not understand some of the contents amounts to 62.9%. This figure is 59.6% among those using GEAI based on company instruction, and 71.8% among those using GEAI on their own initiative. Moreover, while only 1.2% of GEAI users report that they hardly comply with these guidelines, 36.3% report that they do not comply with some of the contents. These figures refer to GEAI users for whom such guidelines have already been established and who demonstrate at least a certain level of understanding. Among those using GEAI based on company instruction, the proportion is 35.5%, while among those using GEAI on their own initiative, it rises to 38.4%. These figures indicate that Japanese companies need to actively promote employee understanding and compliance with internal rules and guidelines. To achieve this, it is effective to make use of various opportunities suited to the organisational context – such as briefings, e‑learning programmes, or OJT.
Percentage of GEAI users, by whether employees use GEAI under company’s instruction or on their initiatives
Note: Employees for whom internal rules or guidelines to work appropriately with GEAI have been created were asked: “To what extent do you understand your company's internal policies and guidelines on the use of generative AI in your work?” “To what extent do you comply with your company's internal rules or guidelines on the use of generative AI?”
Source: JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
When asked what initiatives companies and government should take in the next 10 years to reap the positive effects and curb the negative impacts of AI or generative AI, Japanese workers say improving the safety, reliability, and transparency of AI technologies themselves is the most important issue – even more so than the development of labour policies related to AI (Figure 4.18). This is followed by the need to promote technological development that contributes to improving workers’ performance and wages, enhancing working styles and environments, and boosting job opportunities. These findings show Japanese workers feel strongly that the risks arising from the use of AI technologies in the workplace. However, compared to surveyed countries, Japanese AI users are less likely to explicitly say they trust their employer uses only safe and trustworthy AI (Figure 4.19).
Percentage of all employees
Note: All employees were asked: “Thinking about the next 10 years, what initiatives by companies and government do you think will be important for employees to reap the positive effects and curb the negative effects of AI or generative AI?” Respondents could select multiple answers.
Source: JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
In the finance and insurance sector, the proportion of AI users who trust that their employer uses only safe and trustworthy AI is highest in the United Kingdom at 85.2%, whereas in Japan it is 60.8%, the lowest among the surveyed countries. In the manufacturing sector, this proportion is highest in Ireland at 88.5%, while Japan ranks second lowest among the surveyed countries at 70.6%. That said, across all industries in Japan, the proportion of AI users expressing trust stands at 65.9%. The OECD’s earlier survey and the JILPT survey both also asked AI non-users whether they would trust or not trust their company to use only safe and trustworthy AI, assuming that their company were to adopt AI in the workplace. These results indicate that, compared to AI non-users in other surveyed countries, AI non-users in Japan are significantly less likely than those in other countries to trust their company to use only safe and trustworthy AI (Annex Figure 4.A.3).
Percentage of AI users, by country
Note: AI users were asked: “To what extent would you trust your company to only use AI that is safe and trustworthy? (Trust completely; Trust somewhat; Do not trust very much; Do not trust at all; I don’t know)”
Source: OECD worker survey on the impact of AI on the workplace (2022), JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
Furthermore, the results of the generalised ordered logit model indicate that there are several statistically significant differences – depending on Japanese workers’ attributes – in their likelihood of trusting or not trusting that their company uses only safe and trustworthy AI (Annex Table 4.A.3, Annex Table 4.A.4). The regression results indicate that AI non-users are generally less likely to trust that their employer uses only safe and trustworthy AI technologies. This tendency is particularly pronounced among non-users who are male, middle‑aged or older, employed by SMEs or companies facing labour shortages, engaged in low-skilled occupations, or residing in specific regions such as the Kinki and Chugoku/Shikoku areas. Although such disparities are less evident among AI users than among non-users, the findings imply that initiatives to foster trust in employers’ use of only safe and trustworthy AI technologies should also target middle‑aged and older AI users, as well as those in low-skilled occupations.
The data supporting these findings are shown below. AI non-users in Japanese are 27.4 p.p. more likely to report that they do not trust their company to use only safe and trustworthy AI, compared to Japanese AI users. Regarding gender, among all Japanese workers, males tend to be more likely to report that they do not trust their company to use only safe and trustworthy AI. However, no statistically significant difference was observed among Japanese AI users. This suggests that the higher likelihood of reporting distrust is primarily driven by males who are not AI users. Regarding age groups, among all Japanese workers, middle‑aged and older workers are clearly more likely to report that they do not trust their company to use only safe and trustworthy AI, compared to younger workers. A similar pattern is observed among Japanese AI users: compared to younger AI users, middle‑aged AI users are 2.8 p.p. more likely to report “Do not trust at all”, and older AI users are 9.9 p.p. less likely to report “Strongly trust”. Regarding educational background, among all Japanese workers, those with a university degree are more likely to report trusting their company to use only safe and trustworthy AI. However, no statistically significant differences were observed among AI users, indicating that this tendency is primarily found among AI non-users with a university degree. Regarding average weekly working hours (including overtime), among all Japanese workers, those with longer working hours tend to be less likely to report “Somewhat trust” and more likely to report “Do not trust at all”. Regarding income levels in 2023, among all Japanese workers, no statistically significant differences were observed for the low-income group compared to the middle‑income group. However, those in the high-income group were less likely to report “Do not trust at all” and more likely to report “Somewhat trust”. Regarding company size, among all Japanese workers, those employed at larger companies were less likely to report “Do not trust at all” and more likely to report trust compared to those at smaller companies. However, no statistically significant differences were observed among AI users, suggesting that this pattern is specific to non-AI users. These company size – based differences in trust may be a distinctive feature observed only in Japan (Annex Figure 4.A.4). Regarding company manpower status, among all Japanese workers, those who reported that their company was experiencing a labour shortage were clearly less likely to report trusting their company to use only safe and trustworthy AI, and more likely to report distrust, compared to those who reported that manpower was appropriate. However, no statistically significant differences were observed among AI users, suggesting that this is a characteristic specific to non-AI users. Regarding occupation, among all Japanese workers, compared to Clerical support workers, those in Plant and machine operators, and assemblers or Elementary occupations were 6 p.p. more likely to report “Do not trust at all”. In contrast, Professionals and Technicians and associate professionals were 1.8 p.p. more likely to report “Strongly trust”. Among Japanese AI users, compared to Clerical support workers, there is a weak but observable tendency for those in groups6 including Elementary occupations to be less likely to report “Somewhat trust” and more likely to report “Do not trust at all”. Regarding industry sector, among all Japanese workers, compared to those in Wholesale and retail trade, no industries showed a statistically significant increase in the likelihood of reporting a lack of trust. However, workers in Construction, Information and communications, Finance and insurance, and Scientific research, professional and technical services exhibited a tendency to be more likely to report that they trust their company. Regarding region, among all Japanese workers, compared to those in Hokkaido/Tohoku, those in the Kinki region were 3.6% more likely to report “Do not trust very much”, while those in the Kinki and Chugoku/Shikoku regions were around 2% less likely to report “Strongly trust”. Among Japanese AI users, no regions showed a statistically significant increase in the likelihood of reporting distrust compared to Hokkaido/Tohoku; however, those in North Kanto/Koshin, South Kanto, Hokuriku/Tokai, and Kinki were less likely to report “Strongly trust” compared to Hokkaido/Tohoku.
Building on the preceding sections, which have highlighted the importance of enhancing the safety and trustworthiness of AI technologies introduced in the workplace in Japan, this final section of the chapter provides an overview of measures taken by various countries to ensure the safety, reliability, and transparency of such technologies. It then turns to a discussion of initiatives implemented in Japan to date.
4.4.1. Bias and discrimination
Through bias and discrimination, AI may undermine inclusiveness in the labour market. Moreover, discrimination may result in an inefficient allocation of resources. While existing anti-discrimination legislation is applicable to AI use in the workplace, there may be gaps and loopholes in this legislation and scholars have argued that the law may need to evolve (Adams‐Prassl, Binns and Kelly‐Lyth, 2022[3]). While relevant case law is still limited, automated decision making systems are increasingly facing challenges (Adams‐Prassl, Binns and Kelly‐Lyth, 2022[3]). In the United States, for example, the Equal Employment Opportunity Commission sued and settled with iTutorGroup after it was found that the company’s recruitment software automatically rejected female applicants aged 55 and over and male applicants aged 60 and older. The bias was discovered when one of the applicants reapplied for the same job using a more recent birthdate and did get an interview (Wilkinson and Lukens, 2023[4]).
The policy response so far contains a mixture of: awareness raising, enforcement of existing legislation, impact assessments/audits, as well as data governance requirements, and (proposed) revisions to the legislation (e.g. by inverting the burden of proof). Fighting bias and discrimination will also depend on transparency and explainability of AI tools (discussed further down).
In the United States, some states are proposing modifications to their anti-discrimination law, including California, where there is a proposal by the California Civil Rights Council to hold employers liable for the use of AI-Based Employment Decision Tools that have a discriminatory impact against an applicant or employee based on a protected characteristic (Wilkinson and Lukens, 2023[4]). At the same time, California introduced AB 331 on 30 January 2023, which would prohibit employers from using automated decision tools in a way that contributes to algorithmic discrimination, and would also, amongst others, require them to perform impact assessments for automated decision tools in use, provide individuals with notice of the use of such tools and allow them to request an alternative process (Wilkinson and Lukens, 2023[4]). On 17 May 2024, Colorado enacted the first comprehensive AI legislation in the United States with a specific focus on discrimination. Developers and deployers must use reasonable care to avoid discrimination via AI systems that make, or are a substantial factor in making, a consequential decision “that has a material legal or similarly significant effect on the provision or denial to any consumer of, or the cost or terms of employment or an employment opportunity”.
However, implementing such laws can be tricky. One of the first attempts at addressing bias in automated employment decision tools was New York City’s Local Law 144 of 2021 which prohibits employers and employment agencies from using such tools unless they have been subject to a bias audit within one year of their use, information about the bias audit is publicly available, and certain notices have been provided to employees or job candidates. Six months after the implementation of the law, researchers found only 18 bias audits and 13 transparency notices from nearly 400 employers analysed (Wright et al., 2024[5]). One of the main reasons for this seemingly low compliance may be that the law narrowly focusses on tools that are used without any human oversight. It is, however, very easy to get round that by arguing there is a human in the loop somewhere (Weber, 2024[6]).
In Canada, while the Canadian Human Rights Act (CHRA) as well as provincial human rights legislation protect individuals against discrimination, it is felt that AI systems risk causing harm to historically marginalised communities on a large scale if not properly assessed for bias. The Artificial Intelligence and Data Act (AIDA), if passed, would address this risk by requiring businesses to proactively identify, assess and mitigate the risk of bias on grounds prohibited by the CHRA prior to a high-impact system being made available for use.
The EU AI Act calls for AI systems to be “developed and used in a way that includes diverse actors and promotes equal access, gender equality and cultural diversity, while avoiding discriminatory impacts and unfair biases that are prohibited by Union or national law”. The EU AI Act places emphasis on the quality of data which should be “sufficiently representative, and to the best extent possible free of errors and complete in view of the intended purpose of the system” and that attention should be paid to “the mitigation of possible biases in the datasets” “to ensure that the high-risk AI system performs as intended and safely and it does not become a source of discrimination”. To achieve this, the EU AI Act states that “providers should, exceptionally, to the extent that it is strictly necessary for the purposes of ensuring bias detection and correction in relation to the high-risk AI systems, […] be able to process also special categories of personal data”. The EU AI Act also calls for “appropriate data governance and management practices”, potentially with recourse to “third parties that offer certified compliance services including verification of data governance, data set integrity, and data training, validation and testing practices.” At the same time, and as mentioned above, the EU bans the use of certain systems such as those “intended to be used to detect the emotional state of individuals in situations related to the workplace”.
In Switzerland, there is an acknowledgement that the current legislation, which only prohibits discrimination by state actors, may need revision and there is a call for a “general equal treatment law that covers and sanctions discrimination by private parties, especially, companies, based on specific protected characteristics” and that the difficulty of proving discrimination “could be solved by reversing the burden of proof” (Thouvenin et al., 2021[7]).
Something else countries could do is promote diversity in the AI workforce (i.e. the subset of workers with skills in statistics, computer science and machine learning who could actively develop and maintain AI systems). Today, this workforce is primarily male (Green and Lamby, 2023[8]) and this is likely to impact the design of AI technologies, consciously or unconsciously, potentially resulting in bias and discrimination.
4.4.2. Protecting worker health and safety
Most countries have regulations that set out employers’ obligations towards employees concerning their occupational safety and health. While details vary from country to country, employers usually have to assess risks, eliminate or reduce these risks with preventative and protective measures, and inform workers about the risks and train them. While in theory such regulations should also cover AI, there may be gaps. Also, while most countries have product liability regulations, they likely will need to be adapted to the use of AI systems. Moreover, labour inspectorates may lack the knowledge and/or capacity to address new risks posed by AI.
In the EU, Directive 89/391/EEC of 12 June 1989 introduced measures to encourage improvements in the safety and health of workers at work, including the obligation for employers to assess the occupational health and safety risks, and it laid out general principles of prevention that employers are to implement. Since AI systems potentially have significant impact on the safety and on the physical and mental health of platform workers, this directive would apply to employers using AI. However, the EU has since worked on more AI-specific legislation.
The EU AI Act is “essentially a product safety regulation” (Martens, 2024[9]): it aims to ensure that AI systems used in the EU are safe. AI systems which are judged to be high-risk (which includes AI systems used in employment, worker management and access to self-employment) will need to be assessed before being put on the market and throughout their lifecycle. To this end, technical and organisational measures should be taken, for example by designing and developing appropriate technical solutions to prevent or minimise harmful or otherwise undesirable behaviour. Those technical solutions may include mechanisms enabling the system to safely interrupt its operation (fail-safe plans) in the presence of certain anomalies or when operation takes place outside certain predetermined boundaries.
On top of the EU AI Act, the Platform Work Directive provides extra protections for workers on digital labour platforms in the EU. For example, the directive expects digital labour platforms to carry out an evaluation, at least every two years, of the impact of automated monitoring or decision making systems on the working conditions of platform workers, amongst others. These mandatory evaluations should indicate whether the safeguards of the systems are appropriate to address those risks and digital labour platforms should take appropriate preventive and protective measures. The digital labour platform should make their risk evaluation and the assessment of the mitigating measures available to platform workers, their representatives and the competent authorities.
In October 2024, the U.S. Department of Labor issued “Principles and Best Practices for Developers and Employers” to promote worker well-being (henceforth the DoL’s AI Principles – although these remain very high-level and highlight the importance of worker consultation and transparency, clear governance systems, protecting labour rights, worker upskilling and responsible data use, amongst others. It is not clear to what extent these principles still reflect policies of the new administration.
In addressing AI-related risks to health and safety, countries could strengthen the capacity of labour authorities to supervise and enforce compliance with the law including through effective sanctions and sanctioning procedures for non-compliance, and through building their capacity to make use of AI in a trustworthy manner.
4.4.3. Data protection and privacy
The growing integration of AI in the world of work will likely result in greater collection and analysis of data on workers and job applicants. In some cases, data on workers is collected as a byproduct and is not the main purpose of the AI system. In other cases, AI systems may be specifically designed for measuring productivity or worker surveillance. Data collected may include information such as: worker movements, biometric data, as well as digital activities.
Workers may worry about their privacy when such data is personal. They might also worry that the data are used for purposes other than for which it was intended. Increased monitoring and surveillance could also lead to stress. In addition, AI may be used to infer personal data. As was pointed out in the United States Executive Order 14 110 “on Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence” before it was revoked, “Artificial Intelligence is making it easier to extract, re-identify, link, infer, and act on sensitive information about people’s identities, locations, habits, and desires. Artificial Intelligence’s capabilities in these areas can increase the risk that personal data could be exploited and exposed.”
The protection of workers against privacy risks varies considerably across OECD countries and gaps remain even in those with the strongest protections.
In the EU, the GDPR regulates the use of personal data by AI systems: (i) requiring such AI systems to be developed with a clearly-defined purpose; (ii) allowing the use of personal data on the basis of only 6 legal grounds (consent, compliance with a legal obligation, performance of a contract, completion of a public interest mission, the safeguarding of vital interests, the pursuit of a legitimate interest); (iii) requiring data minimisation;7 (iv) limiting the retention period; (v) requiring transparency; and (vi) respecting the individuals data rights.8
In practice, data protection and privacy rules for workers across Europe are far from “harmonised, consistent, and comprehensive” because GPDR “leaves the issue of employee data protection to be addressed at the Member State level” (Abraha, 2022[10]). There are various reasons why it is hard to regulate workplace privacy through a general privacy law (Abraha, 2022[11]): (i) employers often rely on supposed legitimate interest for monitoring workers, making it harder to decide what is acceptable and what is not; (ii) the power imbalance implicit in the employer-employee relationship casts doubt over the meaning of consent in such settings; (iii) in the employment context, there are collective rights to information, participation and co-determination, and such rights are missing in general data protection regulations. Many EU member countries have no employment-specific provisions in their data protection laws. However, workers’ personal data can also be regulated through other laws (e.g. labour law) and/or collective agreements. Only a small number of countries, such as Cyprus and Malta, do not include any employment-specific provisions in either their respective data protection law or labour law. In some countries (e.g. Austria, Germany, Sweden) social partners play an important role in the introduction/use of new technologies for monitoring purposes, while this role is more limited in others (e.g. Finland). The rules governing consultation with the social partners on this issue may also be laid down by law (e.g. France). As a result, workers’ data protection in Europe is regulated through a mixture of national labour law and collective agreements, in addition to data protection law, and there are differences across countries in the extent to which workers’ data is protected (Abraha, 2022[11]).
Another challenge in the context of AI, is that the GDPR relies on general principles which leave room for interpretation. Such uncertainty combined with the high penalties of non-compliance could discourage many companies (and in particular small and medium-sized ones) from investing in AI. Guidance from data protection bodies (or other competent authorities) would go a long way in addressing such uncertainty (Sartor and Lagioia, 2020[12]). Thus, the successful application of GDPR to AI will depend on what guidance data protection bodies and other competent authorities provide. Appropriate guidance diminishes the cost of legal uncertainty and directs companies to efficient and data protection-compliant solutions (Sartor and Lagioia, 2020[12]).
The EU platform work directive, by virtue of regulating algorithmic management, takes worker privacy a step further by forbidding: (i) the processing of personal data on the emotional or psychological state of the person performing platform work; (ii) the processing of any personal data in relation to private conversations, including exchanges with other persons performing platform work and their representatives; (iii) the collection of any personal data while the person performing platform work is not offering or performing platform work; (iv) processing personal data to predict the exercise of fundamental rights, including the right of association, the right of collective bargaining and action or the right to information and consultation; (v) the processing of any personal data to infer racial or ethnic origin, migration status, political opinions, religious or philosophical beliefs, disability, state of health, including chronic disease or HIV status, the emotional or psychological state, trade union membership, a person’s sex life or sexual orientation; and (vi) the processing of any biometric data of a person performing platform work to establish that person’s identity by comparing that data to stored biometric data of individuals in a database.
Awareness of these issues is growing, and some countries are considering taking steps. For example, in Germany, political discussions have been underway to enact a German national law on employee data protection in addition to the EU GDPR, which could also contain dedicated provisions on AI (White & Case, 2024[13]).
Gaps in protections are greater in other OECD countries.
In the United States, data protection and privacy laws have been described as a “checkerboard of federal and state privacy” (Kerry, 2020[14]) as well as a “patchwork with varying applicability to the employment context” (Gaedt-Sheckter and Maxim Lamm, 2023[15]), leaving “workers in the U.S. virtually unprotected from employers using digital workforce management technologies” (Feng, 2023[16]).
At federal level in the United States, the Electronic Communications Privacy Act (ECPA) prohibits an employer from intentionally intercepting the oral, wire and electronic communications of employees, unless the monitoring is done for a legitimate business reason, or the employer obtained the employee’s consent. In practice, this offers little protection for employees since both express and implied consent suffice, including the inclusion of a disclaimer in an employee handbook or electronic communications policy that explicitly provides notice to employees that the employee has no expectation of privacy in the use of the company’s communications systems and that the company maintains the right to monitor employee communications (Lee et al., 2022[17]).
While various states have passed privacy laws, most of these laws specifically exempt employee and job applicant data (Gaedt-Sheckter and Maxim Lamm, 2023[15]). California is a notable exception. Under the California Privacy Rights Act (CPRA), workers have the right to know when employers are monitoring them and for what purpose. They can access their data and request to correct or delete it. They can know if employers are profiling them or buying data about them, like social media activity. And they can opt out of employers selling their data (Feng, 2023[16]). There are similarities between the CPRA and the GDPR, but there are also important differences such as: the CPRA does not require a legal basis for the processing of personal data (and hence consent is not required to process information under the CPRA in the same way that processing is defined under the GDPR). In addition, the CPRA only applies to companies of a certain size (OneTrust DataGuidance and Newmeyer & Dillion LLP, 2022[18]).
In some cases, policymakers decide to ban some AI technologies to protect privacy. In the United States, for example, various cities have adopted bans on facial recognition technologies. Similarly, the EU AI Act prohibits AI systems that create or expand facial recognition databases through the untargeted scraping of facial images from the internet or CCTV footage.
Not all countries feel like data protection and privacy regulations need to change. In Switzerland, for example, the view is that the processing of personal data by algorithmic systems does not raise any fundamentally new questions and that it therefore seems possible (in principle) to solve the challenges for the protection of privacy and data using existing data protection law (Thouvenin et al., 2021[7]). Similarly, in New Zealand, the existing Privacy Act is deemed to cover AI tools and the Office of the Privacy Commissioner issued guidance on compliance with privacy law when using AI (New Zealand Privacy Commissioner, 2023[19]).
4.4.4. Human agency and oversight
As algorithms and automation permeate the workplace, there is a risk that individuals lose the ability to control their choices and decisions (i.e. their autonomy). Not only could this result in breaches of human rights and democratic values, but it could also undermine human dignity if it “instrumentalises” individuals or robs them of their sense of self-worth (Teo, 2023[20]).
A particular threat to agency and dignity comes from automated decision making and algorithmic management practices, which gained prominence in the platform economy but are very common in more traditional workplaces too (Milanez, Lemmens and Ruggiu, 2025[21]). The prevalence of algorithmic management tools ranges from 90% of firms in the United States to 79% in European countries and 40% in Japan. These practices involve the use of automated systems to replace functions that managers usually perform, such as allocating tasks, determining work schedules, giving instructions, monitoring and evaluating workers, as well as providing incentives or imposing sanctions. Algorithmic management can be intrusive and can restrict workers’ autonomy and diminish opportunities for human interaction, with potentially important consequences for working conditions and equal treatment at work, amongst others. It could also lead to a shift in management and power relations within the workplace.
While “a total prohibition of automated decision-making would not be feasible” (Lukács and Váradi, 2023[22]), policymakers across OECD countries are taking steps to regulate it by: clarifying the circumstances under which automated decision making may or may not be permitted, requiring human oversight throughout the AI lifecycle as well as transparency (see next section); requiring periodic testing and validation of the AI; and giving individuals rights to: human intervention, express their opinion, to obtain an explanation, and challenge/contest decisions taken by algorithms.
In the EU, automated decision making is regulated by the GDPR. More specifically, Article 22 of the GDPR states that “data subjects shall have the right not to be subject to a decision based solely on automated processing”. In practice, however, very few decisions in the employment context rely solely on automated processing. It is easy to argue that there is a “human in the loop” even if that only means the rubberstamping of AI decisions. There are also exceptions that would allow for decisions based solely on automated processing: (i) if the automated processing is necessary for entering into, or performance of a contract; (ii) when the decision based on automated processing is authorised by “Union or Member State law”; and (iii) if the data subject has given his or her explicit consent. That does not mean there are no requirements for the employer who uses those exceptions to justify decisions based solely on automated processing: individuals subject to such decisions would still have a right to obtain human intervention, to express their point of view, and to contest the decision.
The EU Platform Work Directive takes these regulations a step further. It ensures that a person performing platform work cannot be fired or dismissed based on a decision taken by an algorithm or an automated decision making system. Instead, platforms must ensure human oversight on important decisions that directly affect the persons performing platform work. The data subject has a right to obtain human intervention, to express his or her point of view, to obtain an explanation and to contest the decision. The Directive also states that digital labour platforms shall not use automated monitoring and decision making systems in any manner that puts undue pressure on platform workers or otherwise puts at risk the physical and mental health of platform workers. Digital labour platforms are expected to regularly monitor and evaluate the impact of individual decisions taken or supported by automated monitoring and decision making systems on working conditions. They are expected to tackle systematic shortcomings on the use of automated monitoring and decision making systems. When the outcome of the oversight activities identifies high risks of discrimination at work, or the infringement of rights of persons performing platform work, digital labour platforms should take appropriate measures to address them, including the possibility to discontinuing such systems.
The EU Platform Work Directive also takes steps to ensure that human oversight is meaningful. In particular, digital labour platforms should ensure sufficient human resources for human oversight and the persons charged with the function of overview should have the necessary competence, training and authority to exercise that function and the right to override automated decisions. They should be protected from dismissal, disciplinary measures or other adverse treatment for exercising their functions.
In Australia, the government has agreed with the recommendations made by the Privacy Act Review Report around automated decision making (Australian Government, 2023[23]), including: a requirement for privacy policies to set out the types of personal information that will be used in substantially automated decisions; the introduction of a right for individuals to request meaningful information about how substantially automated decisions are made. These recommendations are very much in line with the EU GDPR except that they extend to “substantially automated” decisions, rather than being restricted to “solely automated decisions” because “few decisions are made without any level of human intervention, and if the proposal was restricted to solely automated decisions, entities could potentially bypass requirements by including a negligible level of human involvement” (Australian Government Attorney-General’s Department, 2022[24]).
In Canada there has been a Directive on Automated Decision-Making since 2019, however it only applies to tools used in administrative decision making. The directive requires algorithmic impact assessments prior to the production of an automated decision system, as well as the regular reviewing and updating of it. Other important aspects of the directive include: transparency (e.g. providing notice before decisions, or explanations after decisions) as well as recourse for clients wishing to challenge a decision.
In the United States, a proposal at federal level (the “No Robot Bosses Act”) had sought to prohibit certain uses of automated decision systems (ADS) by employers, require employers to disclose how and when ADS are being used, and to add protections for employees and applicants related to ADS. In particular, the act would have: (i) prohibited employers from relying solely on ADS in making employment-related decisions; (ii) required pre‑deployment and periodic testing and validation of ADS for issues like discrimination or bias before they are used in employment-related decisions; (iii) required employers to train individuals or entities on the operation of ADS; (iv) directed employers to provide independent human oversight of ADS outputs before using outputs in employment-related decisions; (v) required employers to timely disclose their use of ADS, the data inputs and outputs of these systems, and employee rights related to decisions aided by these systems; and (vi) established a Technology and Worker Protection Division at the Department of Labour to regulate the use of ADS in the workplace. However, it is currently unclear whether this act will become law.
At state level in the United States, some legislation around automated decision making has already been adopted, mainly around its use in recruitment, while further legislation is being proposed. As mentioned above, New York City Local Law 144 prohibits employers and employment agencies from using Automated Employment Decision Tools (AEDTs) unless they meet certain requirements around bias and transparency.
4.4.5. Explainability
AI systems, particularly those using complex technologies like deep neural networks, yield outcomes that can be difficult or even impossible to explain. A lack of explainability can undermine the trust and confidence that people place in AI systems and the decisions that are informed by them. It also makes it difficult for individuals to provide informed consent to the use of such systems, or to identify and seek redress for adverse effects caused by AI systems in the workplace. A lack of trust and confidence, in turn, can cause worker resistance and hence hinder the adoption of AI systems in the workplace.
The platform economy is replete with examples of workers who faced “black box” algorithms and did not know the reasons for decisions taken by automated systems, while at the same time lacking the possibility to obtain an explanation, to question and discuss those decisions with a human being, or to contest them and seek rectification and redress.
The EU Platform Work Directive seeks to address this. It states that “persons performing platform work have the right to obtain an explanation from the digital labour platform for any decision taken or supported by an automated decision-making system without undue delay. The explanation, in oral or written form, shall be presented in a transparent and intelligible manner, using clear and plain language.” In practical terms, the digital labour platform needs to “provide persons performing platform work with access to a contact person designated by the digital labour platform to discuss and to clarify the facts, circumstances and reasons having led to the decision. Digital labour platforms shall ensure that such contact persons have the necessary competence, training and authority to exercise that function.” Persons performing platform work will also “have the right to request the digital labour platform to review the decisions”.
While the Platform Work Directive covers only platform workers, the EU AI Act has extended similar provisions to all workers who interact with AI systems. More specifically, the EU AI Act states that “affected persons should have the right to request an explanation when a decision is taken by the deployer with the output from certain high-risk systems […] which produces legal effects or similarly significantly affects him or her in a way that they consider to adversely impact their health, safety or fundamental rights. This explanation should be a clear and meaningful and should provide a basis for affected persons to exercise their rights”. In a way, this is an extension of the GDPR which already required that data subjects be provided with “meaningful information about the logic involved” in automated decision making processes.
In Canada, amendments to the Quebec Act that entered into force in September 2023 now regulate automated decision making based on the processing of personal information, and require disclosure of the occurrence of processing and the provision of information about the reasons and factors that have led to a decision (White & Case, 2024[13]).
4.4.6. Accountability
While rules to ensure safety will reduce risks, they do not eliminate those risks entirely and, when they materialise, there may be harm or damage. In such cases, it is important to determine who is liable, however this is made particularly difficult in the case of AI given the complexity, autonomy and opacity of such tools, which can make it difficult or expensive for victims to identify the liable person and prove the requirements for a successful liability claim, and which may deter them from claiming compensation altogether.
To address this, the EU was preparing Liability Rules for Artificial Intelligence which would have eased the burden of proof through the use of disclosure and rebuttable presumptions. It would have established for those seeking compensation for damage a possibility to obtain information on high-risk AI systems to be recorded/documented pursuant to the AI Act. In addition to this, the rebuttable presumptions would have given those seeking compensation for damage caused by AI systems a more reasonable burden of proof and a chance to succeed with justified liability claims. The Liability Rules for AI have, however, been abandoned.
In Switzerland, it is felt that “the norms of general liability law also apply to [algorithmic] systems” (Thouvenin et al., 2021[7]) and that “in certain sectors, strict liability rules that apply to algorithmic systems (e.g. for vehicles in the Road Traffic Act or drones in the Air Traffic Act) are already available” (Thouvenin et al., 2021[7]), but also there is a question as to whether “strict operator liability should be introduced for operators of algorithmic systems in other sectors” and whether “the Swiss Product Liability Act must be updated” (Thouvenin et al., 2021[7]).
According to the Canada AIDA, accountability “means that organisations must put in place governance mechanisms needed to ensure compliance with all legal obligations of high-impact AI systems in the context in which they will be used. This includes the proactive documentation of policies, processes, and measures implemented.”
In recent years, legislators across the OECD have made efforts to promote accountability mechanisms, such as requiring impact assessments and/or audits of AI systems to provide evidence and assurance that they are trustworthy and safe to use.
In Canada, the AIDA, if passed, would introduce obligations to establish and maintain written accountability frameworks by persons who make a general-purpose system or high-impact system available, or who manage the operations of such systems. Written accountability frameworks would include: (i) a description of the roles, responsibilities, and reporting structure for all personnel who contribute to making the system available or managing its operations; (ii) policies and procedures respecting the management of risks relating to the system and the data used by the system; and (iii) anything else required by regulation (White & Case, 2024[13]).
In the United States, New York City Local Law 144 expects employers using automated employment decisions tools to complete yearly bias audits.
The EU AI Act requires providers to carry out ex ante conformity assessments of high-risk AI tools before they are placed on the market, and to put in place post-market monitoring systems. The EU AI Act also places accountability with deployers (including employers) to use AI systems in accordance with the instructions of use. Indeed: “Whilst risks related to AI systems can result from the way such systems are designed, risks can as well stem from how such AI systems are used.”
Accountability is closely linked to transparency and the information about AI systems that permit traceability – including technical documentation which demonstrates the compliance of the AI system with relevant requirements, as well as the automatic recording of events (logs) over the duration of the lifetime of the system (EU AI Act).
Accountability is also linked to human oversight, to ensure that AI systems are used as intended and that their impacts are addressed over the system’s lifecycle. This includes making sure the system is subject to in-built operational constraints that cannot be overridden by the system itself and are responsive to the human operator, and that the natural persons to whom human oversight has been assigned have the necessary competence, training and authority to carry out that role.
Finally, skills and AI literacy are key to promoting accountability. The EU AI Act states that “AI literacy should equip providers, deployers and affected persons with the necessary notions to make informed decisions regarding AI systems”. In the United States, the Federal Government has announced that it will work to ensure that all members of its workforce receive adequate training to understand the benefits, risks, and limitations of AI for their job functions (BEO).
In 2023, under Japan’s presidency, the G7 launched the “Hiroshima AI Process”9 to consider global rules related to generative AI, and established “Hiroshima Process International Guiding Principles for All AI Actors” and a “Hiroshima Process International Code of Conduct for Organizations Developing Advanced AI Systems” (hereinafter referred to as “Hiroshima Process International Guiding Principles, etc.”) aimed at realising safe, secure, and trustworthy AI. Subsequently, Japan has been actively engaging in outreach beyond the G7 by launching the Hiroshima AI Process Friends Group, working on expanding the number of countries and regions that support the spirit of the Hiroshima AI Process.
On the other hand, within Japan, efforts to address risks and issues arising from AI have been guided by a policy approach that seeks to balance the benefits of AI – such as promoting innovation, enhancing productivity, and alleviating labour shortages – with the need to mitigate associated risks. To avoid excessive regulation that might hinder Japan’s international competitiveness, the government has primarily respected the autonomy of AI developers and users, while combining existing legislation with soft law instruments such as guidelines and official interpretations issued by relevant ministries and agencies. For example, in April 2024, the MIC and the METI published the “AI Guidelines for Business Ver1.0”,10 which state that AI systems and services should be developed, provided, and used respecting the rule of law, human rights, democracy, diversity, and a fair and just society. Additionally, in June 2023, the Personal Information Protection Commission (PPC) issued notices on the use of GEAI services. In these notices, the PPC compiled the points for businesses and administrative entities to be aware of when handling personal information and calls for the proper handling of personal information, in accordance with the Act on the Protection of Personal Information (APPI). At the same time, the PPC issued notices indicating the points for general users to be aware of when handling personal information in the use of GEAI services. Furthermore, in March 2024, the Agency for Cultural Affairs presented a certain way of thinking related to the interpretation of the relationship between AI and the current Copyright Act. Similarly, in May 2024, the Cabinet Office presented a certain way of thinking related to the interpretation of the relationship between AI and intellectual property rights. Moreover, in February 2024, following similar developments in the United Kingdom and the United States, Japan established its own AI Safety Institute (AISI). The AISI is engaged in the development and promotion of evaluation methods and standards related to AI safety, with the aim of ensuring the deployment of AI systems that are safe, secure, and trustworthy.11
4.5.1. The evolution of AI-related legislation in Japan
As of May 2025, a notable development in Japan is the submission of the “Bill on the Promotion of Research, Development and Utilisation of Artificial Intelligence‑Related Technologies12“ (commonly referred to as the Bill on AI Act) to the 217th session of the National Diet. Following deliberations during the session, the bill was passed and enacted on 28 May. This marks the establishment of Japan’s first domestic legislation specifically aimed at the promotion of AI development and utilisation.
In the interim report compiled on 4 February 2025 by the Cabinet Office’s AI Strategy Council and the AI Institutional Study Group – which had conducted discussions leading up to the submission of the bill – the following general direction was outlined: “AI, including generative AI, is highly versatile and is used in various fields, and there are different ways to mitigate risks. With regard to mitigating risks including the handling of personal information or copyright and dealing with disinformation and misinformation, the basic approach is to respond mainly through existing laws, etc. However, in the case of AI, there are also instances where a cross-cutting response is required. Therefore, it is necessary to strengthen the government’s function as a strategic leadership body which oversees the overall landscape, to formulate strategies, to ensure transparency and appropriateness in order to improve safety, and, where necessary, it is appropriate to develop legal frameworks.”
In line with this direction, Japan’s newly enacted AI Act consists of (1) fundamental principles concerning the promotion of AI development and the assurance of its safety, (2) the responsibilities of relevant stakeholders, (3) key policy measures, (4) the formulation of AI Basic Plan by the government, (5) the establishment of AI Strategic Headquarters within the Cabinet, and (6) supplementary provisions (for more detailed information, see Annex 4.B.).
While the specific timeline for the implementation of policies related to the strengthening of the Japanese Government’s overarching strategic leadership function under the new AI legislation remains unclear at the time of writing, discussions within the government are expected to advance in the near future. One particularly notable feature of the new AI Act – especially in relation to ensuring the safety and trustworthiness of AI – is the explicit legal establishment of the government’s authority to investigate and analyse serious incidents involving infringements of individuals’ rights or interests, such as criminal misuse, personal data breaches, or copyright infringements. Under this provision, AI-utilising businesses shall endeavour to co‑operate with these investigations. Based on the findings, the government may provide guidance, advice or information if necessary. It is expected that the combination of various measures introduced under the new AI legislation will not only help deter the occurrence of inappropriate incidents arising from the use of AI, but also enable the implementation of effective measures to prevent the escalation of harm should such incidents occur. Moreover, although the legislation does not include specific penal provisions, it seeks to address violations through the application of existing laws – such as the Penal Code, the Act on the Protection of Personal Information (APPI), and the Copyright Act – in order to avoid hindering AI innovation through excessive regulation.
4.5.2. The ongoing discussions regarding the APPI in Japan
Under Japan’s new AI legislation, the Japanese Government clarified that it has the legal authority to investigate and analyse serious issues arising from the use of AI, such as those affecting the labour market. However, any penalties applicable to such issues are to be addressed within the framework of existing laws, such as labour-related legislation and the APPI. Currently, when acquiring “sensitive personal information”,13 businesses handling personal information are required, in principle, to obtain consent from the individuals concerned. Some companies have argued this requirement imposes a burden on them, because they have to verify whether their datasets contain any such information and, if so, either obtain the necessary consent or delete the data. Given these and other circumstances, discussions have emerged around whether such consent should be waived when the data is used solely for statistical purposes, such as developing and training AI systems to generate statistics. Debates are expected to continue, considering diverse views on how best to strike a balance between promoting the use of AI technologies and addressing AI-related risks, and the Japanese Government aims to submit the relevant bill at an early date.
4.5.3. The MHLW’s efforts so far and recommendations for future action
Japan is pursuing a policy of actively promoting the use of AI technologies in the workplace. The Basic Policy on Economic and Fiscal Management and Reform 2025, approved by the Cabinet, underscores that AI and digital technologies should be utilised to address structural challenges such as labour shortages and population ageing. The strategy also highlights plans to expand the range of training courses eligible for ETB in the area of digital skills, including AI, and to enable non-regular workers across the country to access vocational training online. Furthermore, within the MHLW, the Subcommittee on Basic Labour Policy of the Labour Policy Council published a report14 in May 2025, highlighting the importance of actively promoting the use of AI to resolve issues faced by SMEs and companies in regional areas. In addition, MHLW conducted research into the use of advanced technologies – such as AI and the metaverse – in the field of human resources, with the aim of identifying and addressing legal and regulatory challenges within the labour framework.15
On the other hand, there are currently no concrete moves to amend labour-related legislation specifically in response to risks arising from the use of AI in the Japanese workplace. That said, there are areas that can be effectively dealt with through existing legal provisions. If a malfunction of an AI-equipped autonomous machine results in harm to the life or physical safety of employees, it may constitute a violation of the Industrial Safety and Health Act, potentially leading to the application of penalties. In addition, if AI technologies that lack reliability and safety are used in employment-related decisions – such as recruitment, placement services, dismissal, wage determination, or performance evaluation – and such use results in discriminatory bias, it may violate provisions of the Employment Security Act or the Labour Standards Act, which could also give rise to legal sanctions. In Japan, social dialogue concerning the introduction of AI into the workplace is fundamentally governed by the principle of autonomy between labour and management. However, if a worker union requests explanations or disclosure of information regarding AI technologies introduced in the workplace, but their employer persistently fails to respond adequately, such conduct may be considered a refusal to bargain in good faith, as prohibited under Article 7 of the Trade Union Act. In such cases, the union may file for dispute resolution with the Labour relations commission,16 which is established under the same Act.
As highlighted in this report, Japan has a lower rate of AI use in the workplace than other countries, and even among AI users, the perceived positive impacts of AI are less pronounced. At the same time, Japanese workers show particularly low levels of trust that their employers are using safe and trustworthy AI technologies. In this context, MHLW policymakers should vigorously continue to promote AI-related labour market policies with a dual focus: boost the use of AI while simultaneously addressing the associated risks. Specifically, MHLW should take a proactive approach in clarifying interpretations and policy positions within the existing legal framework, as a form of soft law, where such clarification is considered conducive to facilitating the use of AI in the workplace and managing associated risks. In addition, MHLW should collect good practices from companies that have successfully realised both the benefits of AI use in the workplace and effective responses to AI-related risks (e.g. third-party evaluations of AI technologies used in the workplace), and share this information – together with relevant data and evidence on workplace AI use – in a manner that is easily accessible to a wide audience. When AI adoption in Japan has increased and sufficient knowledge has been accumulated within MHLW regarding AI use in the workplace, MHLW policymakers could also consider developing guidelines to support both employers and workers in addressing the benefits and risks associated with AI use in the workplace. The guidelines could be regularly updated as the to reflect new technological developments.
4.5.4. Enhancing public employment services through the use of AI
The new AI legislation stipulates that “the government shall actively promote the use of AI technologies within national administrative agencies in order to enhance the efficiency and sophistication of administrative functions.” Although initiated prior to the enactment of the new AI legislation, the initiative by the MHLW in September 2024 to establish a project team to explore the utilisation of AI technologies in Japan’s public employment services (PES), Hello Work, is aligned with the legislation’s fundamental policy direction. A report summarising the discussions of the team was published by MHLW in April 2025.17
During fiscal year 2025, MHLW plans to conduct a pilot programme at ten Hello Work offices across Japan to evaluate the effectiveness and challenges of using AI technologies in specific duties performed by PES staff, including: recommending job vacancies to jobseekers and proposing adjustments to job requirements to employers. In this initiative, AI-generated outputs are to be used solely as reference materials for PES staff decision making and will not be directly communicated to jobseekers or employers. Additionally, in fiscal year 2025, MHLW will implement a pilot test of a generative AI-powered “concierge function” added to its internet services that can be used for job vacancy searches and various employment insurance procedures provided through Hello Work. This function is expected to use a chatbot to respond to questions about how to conduct a job search, guide users to relevant websites, and direct them to Hello Work offices as necessary. As this remains a pilot initiative, the GEAI-based functions will be made available only to users who have agreed to the terms of use, which include a warning regarding the input of personal data.
Building on the outcomes of the two pilot projects and drawing on insights from the OECD Secretariat regarding the utilisation of AI technologies in PES in other countries, discussions will continue within the MHLW from FY2 026 onwards to achieve both the streamlining and advancement of administrative services through the use of AI technologies, and the assurance of their safety and trustworthiness.
Percentage of AI users, by country
Note: AI users were asked: “Has your company provided or funded training so that you can work with AI? (Yes; No; Don’t know)” The figure shows the proportion of AI users who responded “Yes”.
Source: OECD worker survey on the impact of AI on the workplace (2022), JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
Note: Estimates are based on 1 773 observations (AI users). Controls include gender, age group, educational background, employment status, company size, labour shortages or excesses, occupation. Due to a violation of the proportional odds assumption (Brant test), a generalised ordered logit model was employed. *** Significant at the 1% level, ** 5% level, * 10% level.
Source: JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
Note: Estimates are based on 1 773 observations (AI users). Controls include gender, age group, educational background, employment status, company size, occupation, residential area. Due to a violation of the proportional odds assumption (Brant test), a generalised ordered logit model was employed. *** Significant at the 1% level, ** 5% level, * 10% level.
Source: JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
Percentage of AI users, by country
Note: Al users were asked: “In your experience, does your employer consult workers or worker representatives regarding the use of new technologies in the workplace?"
Source: OECD worker survey on the impact of AI on the workplace (2022), JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
Note: Estimates are based on 13,518 observations. Based on the estimation results of a generalised ordered logit model, in which the dependent variable consists of four categories of responses regarding whether respondents trust or do not trust that their employer uses only safe and trustworthy AI, and the independent variables include gender, age group, educational background, employment status, average weekly working hours (including overtimes), annual income in 2023, company size, labour shortages or excesses, industry sector, occupation, and residential area, the marginal effects of each attribute were calculated. However, categories of independent variables that did not yield statistically significant results in any of the response categories of the dependent variable have been omitted from the table. "Workers with a university degree" refers to the combined total of those who have graduated from a four-year university and those who have completed graduate school. The annual income figures for 2023 are before the deduction of taxes and social security contributions. Income levels are classified as follows: "Low" refers to less than JPY 2,000,000; "Middle" refers to more than JPY 2,000,000 and less than JPY 8,000,000; and "High" refers to more than JPY 8,000,000. Due to a violation of the proportional odds assumption (Brant test), a generalised ordered logit model was employed. *** Significant at the 1% level, ** 5% level, * 10% level.
Source: JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
Note: Estimates are based on 1,605 observations (AI users). Based on the estimation results of a generalised ordered logit model, in which the dependent variable consists of four categories of responses regarding whether respondents trust or do not trust that their employer uses only safe and trustworthy AI, and the independent variables include gender, age group, educational background, employment status, company size, labour shortages or excesses, occupation, and residential area, the marginal effects of each attribute were calculated. Due to a violation of the proportional odds assumption (Brant test), a generalised ordered logit model was employed. *** Significant at the 1% level, ** 5% level, * 10% level.
Source: JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
% of AI non-users (AI adopters) and AI non-adopters, by country
Note: AI non-users were asked: "Imagine that your company was going to adopt AI. To what extent would you trust your company to only use AI that is safe and trustworthy? (Trust completely; Trust somewhat; Do not trust very much; Do not trust at all; I don’t know)"
Source: OECD worker survey on the impact of AI on the workplace (2022), JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
% of all employees, by company size
Note: All employees were asked: "(Imagine that your company was going to adopt AI) To what extent would you trust your company to only use AI that is safe and trustworthy?" The figure shows the proportion of workers who answered that they trust (completely or somewhat) own company to only use AI that is safe and trustworthy.
Source: OECD worker survey on the impact of AI on the workplace (2022), JILPT worker survey on the impact of the introduction of AI into the workplace on working practices (2024).
The new AI Act in Japan stipulates that:
1. Fundamental principles.
The promotion of the research, development and utilisation of AI technologies18 shall be carried out with the aim of maintaining Japan’s capacity to conduct research and development in AI technologies and enhancing the international competitiveness of industries related to AI.
The promotion of the research, development and utilisation of AI technologies shall be carried out with the aim of advancing such efforts in a comprehensive and systematic manner, in light of the fact that initiatives by relevant parties – from basic research to practical applications in daily life and economic activities – are mutually and closely interconnected.
In order to ensure the proper conduct of the research, development and utilisation of AI technologies, measures shall be taken to secure transparency in the processes of such research, development and utilisation, as well as other related necessary measures.
The research, development and utilisation of AI technologies shall be promoted under international co‑operation, with a view to contributing to the peace and development of Japan and the international community, and efforts shall be made for Japan to play a leading role in international co‑operation concerning the research, development and utilisation of AI technologies.
2. The responsibilities of relevant stakeholders.
The new AI legislation stipulates the responsibilities of the national government, local governments, research and development institutions, AI-utilising businesses, and the public, in accordance with the fundamental principles. In particular, it is stipulated that the national government bears the responsibility to formulate and implement, in a comprehensive and systematic manner, policies for the promotion of the research, development, and utilisation of AI technologies. It is also stipulated that AI-utilising businesses are to endeavour to proactively enhance the efficiency and sophistication of their business activities and to create new industries through the active use of AI technologies, and that they shall endeavour to co‑operate with the policies implemented by the national and local governments. Furthermore, it is stipulated that the public is to endeavour to deepen their understanding of and interest in AI technologies, and to make efforts to co‑operate with the policies implemented by the national and local governments.
3. Key policy measures.
The national government shall promote integrated research and development ranging from basic research on AI technologies to research and development for their practical application, establish frameworks to support the transfer of research outcomes within research and development institutions, provide information related to such outcomes, and implement other relevant measures.
The national government shall take necessary measures to develop and promote the shared use of facilities, equipment, datasets, and other intellectual infrastructure required for large‑scale information processing, telecommunications, storage of electromagnetic records, and other related functions necessary for the research, development, and utilisation of AI technologies, in order to ensure that such infrastructure is broadly accessible to research and development institutions and AI-utilising businesses.
The national government shall take necessary measures, including the development of guidelines in accordance with the intent of international norms and other related measures, in order to ensure the proper conduct of the research, development, and utilisation of AI technologies.
The national government shall take necessary measures to secure, develop, and enhance the capabilities of human resources possessing specialised and broad knowledge across diverse fields, which are required at each stage from basic research on AI technologies to their utilisation in daily life and economic activities, while ensuring close collaboration and co‑operation with relevant stakeholders.
The national government shall implement necessary measures, including the promotion of education and learning and the enhancement of public awareness activities relating to AI technologies, as well as other related initiatives, so that the public may broadly deepen their understanding of and interest in AI technologies.
The national government shall collect information on trends in the research, development and utilisation of AI technologies both domestically and internationally; analyse cases in which the rights and interests of individuals have been infringed as a result of the research, development or utilisation of AI technologies for improper purposes or by inappropriate methods; examine countermeasures based on such analyses; and conduct other surveys and studies that contribute to the promotion of AI-related research, development and utilisation. Based on the findings thereof, the government shall implement necessary measures, including the provision of guidance, advice and information to research and development institutions, AI-utilising businesses, and other relevant parties.
The national government shall promote international co‑operation concerning the research, development and utilisation of AI technologies, and shall actively participate in the formulation of international norms.
4. The formulation of AI Basic Plan by the government.
The government shall, in accordance with the fundamental principles and based on the basic policy measures, formulate a basic plan for the promotion of the research, development and utilisation of AI technologies (“AI Basic Plan”).
5. The establishment of AI Strategic Headquarters within the Cabinet.
In order to comprehensively and systematically promote policies for the research, development and utilisation of AI technologies, AI Strategic Headquarters shall be established within the Cabinet. The Prime Minister shall serve as the head of the Headquarters, the Chief Cabinet Secretary and the minister in charge of AI strategy shall serve as Deputy Heads, and all other Ministers of State shall serve as its members. The Strategic Headquarters for AI may request relevant administrative organisations, local governments, incorporated administrative agencies, and other entities to submit materials, express opinions, provide explanations, or otherwise co‑operate as necessary.
Supplementary provisions.
This Act shall come into force on the date of its promulgation; provided, however, that the provisions set forth in (4) and (5) shall come into force on the date specified by Cabinet Order within a period not exceeding three months from the date of promulgation.
The government shall, considering international trends concerning policies for the promotion of the research, development and utilisation of AI technologies, as well as other changes in socio‑economic conditions, review the status of the enforcement of this Act, and, when it deems necessary, take appropriate measures based on the results of such review.
References
[11] Abraha, H. (2022), “A pragmatic compromise? The role of Article 88 GDPR in upholding privacy in the workplace”, International Data Privacy Law, Vol. 12/4, pp. 276-296, https://doi.org/10.1093/idpl/ipac015.
[10] Abraha, H. (2022), The GDPR and Privacy in the Workplace, https://blogs.law.ox.ac.uk/oblb/blog-post/2022/09/gdpr-and-privacy-workplace (accessed on 31 May 2024).
[3] Adams‐Prassl, J., R. Binns and A. Kelly‐Lyth (2022), “Directly Discriminatory Algorithms”, The Modern Law Review, Vol. 86/1, pp. 144-175, https://doi.org/10.1111/1468-2230.12759.
[23] Australian Government (2023), Government response to the Privacy Act Review Report, https://www.ag.gov.au/sites/default/files/2023-09/government-response-privacy-act-review-report.PDF (accessed on 11 June 2024).
[24] Australian Government Attorney-General’s Department (2022), Privacy Act Review Report 2022, https://www.ag.gov.au/sites/default/files/2023-02/privacy-act-review-report_0.pdf (accessed on 11 June 2024).
[16] Feng, K. (2023), Overview of New Rights for Workers under the California Consumer Privacy Act, https://laborcenter.berkeley.edu/overview-of-new-rights-for-workers-under-the-california-consumer-privacy-act/ (accessed on 7 June 2024).
[15] Gaedt-Sheckter, C. and E. Maxim Lamm (2023), AI in Employment: Privacy Regulation Is Here, https://www.gibsondunn.com/wp-content/uploads/2023/09/Gaedt-Sheckter_PLI-Chronicle_August_2023.pdf (accessed on 5 June 2024).
[8] Green, A. and L. Lamby (2023), “The supply, demand and characteristics of the AI workforce across OECD countries”, OECD Social, Employment and Migration Working Papers, No. 287, OECD Publishing, Paris, https://doi.org/10.1787/bb17314a-en.
[2] JILPT (2023), JILPT survey on the introduction of digital technologies such as AI and labour-management communication, https://www.jil.go.jp/institute/research/2021/documents/210.pdf (accessed on 19 May 2025).
[14] Kerry, C. (2020), Protecting privacy in an AI-driven world, Brookings Research, https://www.brookings.edu/articles/protecting-privacy-in-an-ai-driven-world/ (accessed on 5 June 2024).
[17] Lee, J. et al. (2022), A Privacy and Employment Law Primer: Recent Updates on Discrimination and Privacy Implications of Technology in the Workplace, Loeb & Loeb, https://www.loeb.com/en/insights/publications/2022/08/recent-updates-on-discrimination-and-privacy-implications-of-technology-in-the-workplace (accessed on 5 June 2024).
[22] Lukács, A. and S. Váradi (2023), “GDPR-compliant AI-based automated decision-making in the world of work”, Computer Law & Security Review, Vol. 50, p. 105848, https://doi.org/10.1016/j.clsr.2023.105848.
[9] Martens, B. (2024), The European Union AI Act: premature or precocious regulation?, Bruegel, https://www.bruegel.org/analysis/european-union-ai-act-premature-or-precocious-regulation (accessed on 30 May 2024).
[21] Milanez, A., A. Lemmens and C. Ruggiu (2025), “Algorithmic management in the workplace: New evidence from an OECD employer survey”, OECD Artificial Intelligence Papers, No. 31, OECD Publishing, Paris, https://doi.org/10.1787/287c13c4-en.
[19] New Zealand Privacy Commissioner (2023), Artificial intelligence and the Information Privacy Principles, https://www.privacy.org.nz/assets/New-order/Resources-/Publications/Guidance-resources/AI-Guidance-Resources-/AI-and-the-Information-Privacy-Principles.pdf (accessed on 16 January 2025).
[1] OECD (2025), Generative AI and the SME Workforce: New Survey Evidence, OECD Publishing, Paris, https://doi.org/10.1787/2d08b99d-en.
[18] OneTrust DataGuidance and Newmeyer & Dillion LLP (2022), Comparing privacy laws: GDPR v. CCPA & CPRA, https://www.dataguidance.com/sites/default/files/gdpr_v_ccpa_and_cpra_v6.pdf (accessed on 7 June 2024).
[12] Sartor, G. and F. Lagioia (2020), The impact of the General Data Protection Regulation (GDPR) on artificial intelligence, European Union, Brussels, https://www.europarl.europa.eu/RegData/etudes/STUD/2020/641530/EPRS_STU(2020)641530_EN.pdf (accessed on 31 May 2024).
[20] Teo, S. (2023), “Human dignity and AI: mapping the contours and utility of human dignity in addressing challenges presented by AI”, Law, Innovation and Technology, Vol. 15/1, pp. 241-279, https://doi.org/10.1080/17579961.2023.2184132.
[7] Thouvenin, F. et al. (2021), A Legal Framework for Artificial Intelligence, University of Zurich, https://www.dsi.uzh.ch/en/research/projects/strategy-lab/strategy-lab-21.html (accessed on 30 May 2024).
[6] Weber, L. (2024), “New York City Passed an AI Hiring Law. So Far, Few Companies Are Following It.”, The Wall Street Journal, https://www.wsj.com/business/new-york-city-passed-an-ai-hiring-law-so-far-few-companies-are-following-it-7e31a5b7 (accessed on 30 May 2024).
[13] White & Case (2024), AI Watch: Global regulatory tracker, https://www.whitecase.com/insight-our-thinking/ai-watch-global-regulatory-tracker (accessed on 16 August 2024).
[4] Wilkinson, C. and D. Lukens (2023), The Growing Regulation of AI-Based Employment Decision Tools, Perkins Coie News & Insights, https://www.perkinscoie.com/en/news-insights/the-growing-regulation-of-ai-based-employment-decision-tools.html (accessed on 30 May 2024).
[5] Wright, L. et al. (2024), “Null Compliance: NYC Local Law 144 and the challenges of algorithm accountability”, The 2024 ACM Conference on Fairness, Accountability, and Transparency, https://doi.org/10.1145/3630106.3658998.
← 1. If the course is commenced by September 2024, up to 70% of the training cost is eligible for reimbursement, with an annual ceiling of 560 000 JPY.
← 2. If the course is commenced by September 2024, up to 40% of the training cost is eligible for reimbursement, with an annual ceiling of 200 000 JPY.
← 3. More precisely, the response options for this question are based on a six‑point scale that includes “Don’t know” as one of the choices.
← 4. The Regional Consortiums for Vocational Abilities Development Promotion is organised by prefectural governments and prefectural labour bureaus and brings together municipalities that operate public vocational training facilities, vocational and educational training providers, workers’ organisations, employer associations, and academic experts.
← 5. The Human Resources Development Subsidies is designed to support employers who are working to improve working conditions and create more attractive workplaces, with the aim of securing and retaining employees. As of July 2025, the subsidy includes seven different programmes. For example, the Employment Management Systems and Workplace Improvement Support Course provides financial assistance for introducing employment management systems or purchasing equipment to reduce employees’ workload.
← 6. This group includes Skilled agricultural, forestry and fishery workers, Craft and related trades workers, Plant and machine operators, and assemblers, Elementary occupations, and Armed forces occupations.
← 7. The personal data collected and used must be appropriate, relevant and limited to what is necessary for the defined objective.
← 8. These rights include access, rectification, erasure, restriction, portability and objection.
← 9. Based on the result of the G7 Hiroshima Summit held in May 2023, this process was established in May 2023 to discuss generative AI, whose rapid development and expansion have become significant issues for the international community as a whole.
← 10. In March 2025, the version 1.1 was released with updates.
← 11. According to news in Japan, the AISI (AI Safety Institute) is developing, in collaboration with private companies, a tool that can automatically assess the safety of AI systems, with plans to make it freely available in August 2025. This tool could make it easier for businesses and individuals to select safe AI technologies for use in their operations. It is designed to check whether AI provides harmful responses – for example, when asked by malicious users about how to make explosives – and whether it generates misinformation. It will also be capable of automatically generating sophisticated prompts intended to cause malfunctions in AI systems in order to test their vulnerabilities.
← 12. It should be noted that, as the official English translation by the Japanese Government has not yet been released, this is a provisional version.
← 13. “Sensitive personal information” in the APPI means personal information as to an identifiable person’s race, creed, social status, medical history, criminal record, the fact of having suffered damage by a crime, or other identifiers or their equivalent prescribed by Cabinet Order as those of requiring special care so as not to cause unjust discrimination, prejudice or other disadvantages to that person.
← 14. Report title; Approaches to ensuring quality employment in regional areas and among SMEs in the context of a rapidly changing society (https://www.mhlw.go.jp/content/12602000/001485054.pdf).
← 15. Survey report on the cutting-edge use of AI and Metaverse technologies in the Human Resources Domain (https://www.mhlw.go.jp/content/11200000/001471931.pdf).
← 16. The Labour relations commission is composed of an equal number of members representing the public benefit, workers, and employers, each serving in a representative capacity.
← 17. Utilising AI at Public Employment Security Offices (Hello Work) with a View to the Future (https://www.mhlw.go.jp/content/11601100/001478507.pdf).
← 18. In the Japanese AI Act, “AI technologies” refers to technologies necessary to realise functions that, through artificial means, substitute for human cognitive, reasoning and decision making capabilities, as well as technologies related to information processing systems that process input data using such technologies and output the results.