Banks and Insurance Companies: AI Is Gaining Momentum, but Governance and Expertise Remain the True Test

Milan, June 10, 2026 – Artificial Intelligence in the Italian financial sector has moved beyond the experimental phase and is entering a new operational phase. This is the finding of a study Cetif the Center for Research on Technologies, Innovation, and Financial ServicesUniversità Cattolica Università del Sacro Cuore—which paints a picture of a rapidly growing market that is still grappling with significant organizational and regulatory challenges.

The figures, compiled by the Cetif Advanced Analytics AI Research Hub and presented today atSummit Cetif Summit, highlight that data is the starting point for transformation. Over 80% of the information held by financial organizations consists of unstructured content, such as emails, contracts, and documents, which has historically been difficult to leverage. AI now makes it possible to unlock the value of this data, transforming it into a tool for decision-making and operational efficiency.

Three stages, one direction: the maturation of AI

• An average of 6.7 machine learning use cases in production per organization: machine learning has now become a well-established foundation for advanced analytics business intelligence;
• More than five average use cases of generative AI currently in testing or at the initial release stage: generative AI is the fastest-growing segment;
• 58% of respondents report organizational challenges in the development of Agentic AI: while Agentic AI has significant potential, barriers to adoption are emerging that relate to skills and internal governance.

An evolution that marks the transition from support solutions to fully-fledged systems capable of managing complex tasks with controlled autonomy.

Applications are becoming increasingly widespread and diverse. AI is finding its place:

• in the automation of document-related processes, reducing time and manual tasks in areas such as onboarding and credit;
• in customer service, with digital assistants capable of handling complex interactions;
• in risk and compliance activities, including anti-money laundering and contract analysis;
• in knowledge management systems, which speed up access to internal information.

Readiness is on the rise, but infrastructure and skills are holding things back

Signs that the market is maturing are also evident in the data:

• The integration of external and unstructured data sources is becoming increasingly important: the use of external data providers is growing significantly, as is the adoption of unstructured data, text, images, and behavioral logs;
• 29% of institutions already make significant use of summary data;
• At the same time, 68% report that infrastructure is not yet adequate to support the full scalability of AI.

Alongside technology, the issue of skills and organizational models is emerging as a key factor, as they are increasingly seen as essential to driving adoption.

Governance: The Real Strategic Issue

The main obstacle to the development of AI in the financial sector is no longer technical, but rather related to governance. The research shows that 47% of institutions view the governance of agentic AI as a separate issue, not linked to existing controls.

The figures confirm this delay:

• only 15% have adopted formal policies on the use of AI;
• only 6% have a dedicated Data & AI Office;
• Although 57% have implemented an ex-ante control framework, only 24% have fully operational second-level controls.

The entry into force of the European AI Act makes this gap even more significant, especially since many use cases—from credit scoringAML—are classified as high-risk.

“Artificial Intelligence in the Italian financial sector has reached a stage of maturity: we are no longer talking about isolated experiments, but about applications that are increasingly integrated into organizations’ core processes,” explains Federico Rajola, Director Cetif Professor at Università Cattolica del Sacro Cuore The real leap forward today is not just technological, but organizational and cultural. The main challenge concerns the ability to effectively govern these solutions, especially in a context where many applications are classified as high-risk by the regulator.”

The research was presented during the Cetif Summit took place today in Milan atUniversità Cattolica Università del Sacro Cuore. The event, which drew nearly a thousand industry professionals and managers (both in person and online), featured presentations from Italy’s leading financial institutions. The event was organized in collaboration with Accenture, DGS, IBM, Reply, and ServiceNow.