Central banks and financial regulators worldwide are falling significantly behind financial institutions in adopting artificial intelligence (AI), raising concerns about their ability to oversee emerging risks linked to advanced AI models. A recent study conducted by the Cambridge Centre for Alternative Finance, in collaboration with the Bank for International Settlements, the International Monetary Fund, and other multilateral organizations, highlights a widening gap between regulators and the financial sector in AI integration and risk management.

Surveying over 350 traditional banks and fintech firms, more than 140 AI vendors, and 130 regulatory entities across 151 countries, the research revealed that financial institutions are adopting AI at more than twice the rate of their supervisors. Only 20 percent of regulators reported having “advanced AI adoption,” while merely 24 percent collect data on AI usage within the industry. Furthermore, 43 percent of regulators indicated they have no plans to begin gathering such information in the next two years.

This lack of data and slow uptake pose challenges to regulators’ ability to monitor and mitigate risks associated with AI, especially as new generation systems like Anthropic’s Mythos emerge. Released earlier this month, Mythos has drawn attention from cybersecurity experts for its potential to exploit software vulnerabilities extensively, which could undermine existing governance and oversight frameworks in banking. The report underscores the difficulty regulators face when holding financial firms accountable for AI-related harms, particularly as more autonomous AI systems provided by third-party vendors gain prominence.

To address these challenges, the study advocates that regulators themselves need to adopt agentic AI technologies—systems capable of acting without human intervention—to keep pace with the evolving landscape they are tasked with supervising. Harish Natarajan, practice manager for competitiveness and innovation at the World Bank, noted that many regulators in emerging markets struggle with inadequate data and skill sets necessary to effectively incorporate AI into their oversight processes.

The report also raises concerns about the financial system’s increasing reliance on a small number of dominant AI providers. It found that 69 percent of all respondents use services from OpenAI, a figure that climbs to 76 percent among financial institutions. This concentration presents significant third-party risks, potentially exposing the global financial sector to vulnerabilities stemming from service disruptions, resilience issues, or price shocks.

As the deployment of AI accelerates in finance, the gap between adoption rates of regulators and the entities they oversee is drawing increased scrutiny, emphasizing the need for greater coordination, data transparency, and technological capability within supervisory bodies.