Anthropic to Brief Financial Stability Board on Mythos AI Security Flaws

Anthropic has agreed to meet with members of the Financial Stability Board to discuss findings from its Mythos AI model. The FSB, a watchdog group comprising finance ministry officials, central bankers, and securities regulators from G20 countries, requested the meeting through Bank of England Governor Andrew Bailey.

The briefing comes as global financial regulators grow increasingly worried that advanced AI models from U.S. tech companies could reveal critical weaknesses in banks' cyber defenses.

Last month, Anthropic disclosed that Mythos had identified thousands of high-severity vulnerabilities across major technology platforms. The AI model found security flaws in every major operating system and web browser it examined.

The company warned that the potential fallout from these vulnerabilities could be severe, with implications for economic stability, public safety, and national security.

Only a handful of companies have received access to Mythos, with most recipients based in the United States. Amazon, Microsoft, and JPMorgan Chase are among the select few granted permission to use the AI model.

Anthropic has maintained limited distribution at the request of the White House, citing security concerns. However, this restricted approach has left many organizations and regulators worried about uneven levels of protection across the financial system.

The Financial Stability Board is preparing a report on sound practices for adopting artificial intelligence in the financial system. The report is scheduled for public consultation in June.

The International Monetary Fund issued a warning this month that AI-driven cyber risk should be treated as a financial stability issue. The IMF noted that such attacks can simultaneously impact payment systems, market confidence, and liquidity.

Artificial intelligence is dramatically reducing the cost and skill level required to conduct cyberattacks at scale. Security experts describe this shift as transforming hacking from a specialized skill into an industrial-scale business.

Google researchers reported what they believe to be the first observed AI-developed zero-day exploit tied to a planned mass exploitation campaign. Security analysts view this incident as evidence of a broader transition toward an industrialized cyber threat environment.