The deployment of artificial intelligence (AI) is increasing in the Italian finance sector, with the insurance and banking sectors leading adoption, and with strong levels of General-Purpose AI (GPAI) development and experimentation. Overall, 39% of the 450 respondents to the OECD project survey report using AI as part of their everyday operations. Among the main financial sectors, insurance has the highest proportion of AI deployment (70% of respondents), followed by banking (59%). The AI deployment rate for financial market players is 31%.
The most commonly reported AI use cases include internal process optimisation and supportive functions that apply across all sectors of finance in Italy. The most frequent purposes include data analysis, text content generation and summarisation. Other common use cases involve anti-money laundering/countering the financing of terrorism (AML/CFT), fraud detection and prevention, as well as customer support (chatbots). Among financial market participants, 60% deploy AI for internal process optimisation, text content generation and translation. The majority of GPAI use cases remain in the development and experimentation phase, indicating a strong exploratory activity.
Despite relatively lower adoption rates, financial market participants are advancing their experimentation with AI and expect to expand its use in core financial market areas. Asset managers reported nearly 1 000 AI use cases in either experimentation or production – second only to banks. Across all sectors, at least half of the companies already using AI are developing or experimenting with additional use cases.
The benefits of AI observed by most respondents span across sectors, including in financial market activity. Three-quarters of companies using AI reported improvements in operational efficiency, while almost two-thirds experienced productivity gains. Many also reported optimisation of internal processes, improved decision-making and the production of new analytical insights. Improvements in areas such as reconciliation efficiencies or management of settlement risk were less common.
Italian firms that deploy AI currently have a strong reliance on the AI services provided by third parties, with strong concentration among the top four providers. Almost 75% of respondents report using third-party cloud services for AI, while 39% rely on GPAI models implemented by a third party, reflecting a strong preference for vendor-supported solutions. At the same time, 39% of firms do not use any free or open-source components, primarily due to security concerns and limited control over data handling.
Firms are adopting heterogeneous approaches to AI governance, with many taking a layered approach that combines multiple governance tools and risk management mechanisms. Sixteen per cent have introduced explicit AI governance frameworks, while others have adjusted existing ones to manage AI risks. Half of the respondents use human oversight (human-in-the-loop) as a key safeguard, noting that most of the applications have limited or no autonomy levels. Accountability for AI outputs is most often allocated to business area users, followed by executive leadership, with just under three-quarters of respondents assigning responsibility to only one function. Almost half of the respondents have not yet implemented safeguards to address AI-specific cyber threats.
Regulatory uncertainty and potential misalignment across rules are the most commonly cited regulatory constraints to wider deployment of AI in finance, with smaller firms particularly affected due to limited resources. Twenty per cent of survey respondents identify a lack of regulatory clarity as a barrier, particularly regarding the implementation of the EU AI Act and its interaction with existing sectorial regulation, including in cross-border activities. Concerns also exist regarding compliance with data protection frameworks, intellectual property rules, regulations on third-party risk and operational resilience.
Firms also reported a range of non-regulatory constraints linked to organisational, skills and cultural factors, data-related challenges, elevated costs, and potential negative impacts of AI outputs. A quarter of firms face challenges in attracting and retaining staff with AI skills, with other constraints including a lack of relevant use cases and limited AI understanding among senior leadership. Data accuracy and consistency are barriers for almost a third of companies, while a quarter face difficulties accessing data and cost constraints. Many also face constraints related to operational and business risks, third-party reliance, limited transparency of third-party AI models, and risk of legal liability or harm to clients.
Italian financial authorities are active in promoting safe and responsible AI development and deployment in the domestic financial sector and within the EU regulatory framework. Italian authorities maintain an active set of measures and tools to monitor AI deployment in finance in Italy. These include supervisory initiatives across all financial authorities in the form of data collection and research, SupTech tools in production and development, and a well-developed ecosystem of innovation facilitators spanning all major segments of financial activity.
A series of policies could help catalyse the responsible adoption of AI in support of more efficient, inclusive and competitive financial markets. More widely, these could also serve to enhance the competitiveness of Italy’s economy while upholding a high standard of consumer protection. Most policy considerations are targeted at Italian financial authorities, including through enhanced co-operation with the non-financial authorities directly concerned. Some are linked to ongoing regulatory initiatives at the EU level, and their implementation will depend on the evolution of the EU legal framework.