As many regulators lag behind financial companies in adopting AI, especially on advanced AI models, the ability of central banks and financial regulators to monitor and counter the risks posed by powerful AI models like Anthropic’s Mythos has been questioned over emerging harms.
Research published Tuesday by the Cambridge Center for Alternative Finance showed that financial institutions are adopting AI at more than twice the rate of their supervisors, with only two in 10 regulators reporting “advanced AI adoption.”
The report found that only 24% of executives surveyed collected data on industry AI adoption, while 43% have no plans to start within the next two years.
“This empirical blind spot could undermine existing optimism (on AI). Authorities cannot successfully exploit or monitor AI if they approach it without solid data on its adoption and risks,” the report said.
The research, produced with the Bank for International Settlements, the International Monetary Fund and other multilateral institutions, included a survey of 350 traditional financial institutions and fintechs, more than 140 AI vendors, and 130 central banks and financial authorities spanning 151 countries.
Role of global regulators monitoring AI:
Regulators and global standard-setting bodies have stepped up warnings about the risks posed by the rollout of AI in the financial sector. Earlier in April, Anthropic released Mythos, which cybersecurity experts saw as posing significant challenges to the banking industry and its legacy technology systems.
Regulators around the world have had conversations with banks about how prepared their legacy systems are for emerging cutting-edge AI models.
The report highlights Mythos as an example of a next-generation system that may soon be able to exploit software vulnerabilities on a large scale, potentially limiting the effectiveness of existing human governance and oversight mechanisms.
“Regulators have generally maintained the principle that financial firms should remain accountable for losses, including cyberattacks, whether the AI is built in-house or supplied by a third party, but that position becomes harder to enforce in the context of more autonomous systems that are provided and managed by third-party vendors,” the authors wrote.
The report says regulators themselves should adopt agentic AI capabilities to match the systems that are able to take action without human oversight.
Harish Natarajan, practice manager of competitiveness and innovation at the World Bank, said at an event to launch the report that officials in emerging market economies often lack the data and skills needed to embed AI.
AI Risk:
The report also raises concerns about the financial sector’s growing reliance on a handful of powerful AI providers.
It found that 69% of all respondents trust OpenAI, rising to 76% among the industry, describing it as a “significantly significant third-party risk consideration” that could expose the global financial system to resilience weaknesses, pricing shocks or supply disruptions.
At the time of the survey, conducted between October 2025 and January 2026, more than half of respondents used Google’s model and slightly more than a third used Anthropic.
