Regulators lag banks in AI as Anthropic’s Mythos raises oversight concerns - BERITAJA

The expertise of cardinal banks and financial regulators to show and combat the risks posed by powerful artificial intelligence models specified arsenic Anthropic’s Mythos has been called into mobility aft a study recovered authorities importantly lag financial firms successful AI take and deficiency information connected emerging harms.

Financial institutions are adopting AI astatine much than doubly the complaint of their supervisors, pinch conscionable 2 successful 10 regulators reporting “advanced AI adoption,” investigation published connected Tuesday by the Cambridge Centre for Alternative Finance showed.

Only 24 per cent of authorities surveyed cod information connected manufacture AI adoption, while 43 per cent person nary plans to commencement wrong the adjacent 2 years, the study found.

“This empirical unsighted spot whitethorn undermine the prevailing optimism [on AI]. Authorities cannot successfully harness aliases oversee AI if they are navigating its take and risks without difficult data,” the study said.

The research, prepared alongside the Bank for International Settlements, the International Monetary Fund and different multilateral institutions, progressive surveying 350 accepted financial institutions and fintechs, much than 140 AI vendors, and 130 cardinal banks and financial authorities spanning 151 countries.

Regulators and world standard‑setting bodies person stepped up warnings about the risks posed by the rollout of AI crossed the financial sector.

Earlier successful April, Anthropic released Mythos, viewed by cybersecurity experts as posing important challenges to the banking manufacture and its bequest exertion systems.

Regulators crossed the globe person engaged pinch banks complete really prepared their bequest systems are for emerging frontier AI models.

The study highlights Mythos as an illustration of next‑generation systems that could soon beryllium could of exploiting package vulnerabilities astatine scale, perchance limiting the effectiveness of existing quality governance and oversight mechanisms.

“Regulators mostly support the rule that financial firms should stay accountable for harms, including cyberattacks, whether AI is built in-house aliases supplied by 3rd parties, but that position becomes harder to use successful the discourse of much autonomous systems that are provided and managed by third-party vendors,” the authors wrote.

The study says regulators must themselves adopt agentic AI capabilities, could of taking actions without quality oversight, to lucifer the systems they oversee.

Harish Natarajan, believe head for competitiveness and invention astatine the World Bank, said astatine an arena to motorboat the study that authorities successful emerging marketplace economies often deficiency the information and skills needed to embed AI.

The study besides flagged concerns about the financial sector’s increasing dependence connected a fistful of powerful AI providers.

It recovered that 69 per cent of each respondents trust connected OpenAI, rising to 76 per cent among the industry, creating what it described arsenic a “notable captious third-party consequence consideration” that could expose the world financial strategy to resilience vulnerabilities, pricing shocks aliases proviso disruptions.

At the clip of the survey, conducted betwixt October 2025 and January 2026, conscionable complete half of respondents utilized Google’s models and a small much than a 3rd utilized Anthropic.