Anthropic is set to brief the Financial Stability Board on cybersecurity vulnerabilities uncovered by its new AI model, Claude Mythos Preview, as global regulators grow more concerned that advanced AI could reshape cyber risk across the financial system.
The briefing follows a request from Bank of England Governor Andrew Bailey, who chairs the FSB. The watchdog coordinates financial rules for G20 economies.
The planned briefing puts Anthropic at the center of a growing debate over whether frontier AI models can help banks find software weaknesses faster — or give attackers the same advantage.
Mythos was announced last month but has not yet been released publicly, and that the model is designed to detect long-standing vulnerabilities in web browsers, infrastructure and software.
Why the FSB is paying attention
The FSB’s interest reflects a broader concern that cyber flaws exposed by advanced AI could become a financial-stability issue, not just a technology problem.
The Financial Times reported that Mythos has identified thousands of severe vulnerabilities across major operating systems and web browsers, and that access to the model is currently limited to about 40 organizations, including Microsoft, Amazon and JPMorgan Chase.
Reuters said cybersecurity experts have warned that Mythos could enable more sophisticated cyberattacks, posing risks for banks that still rely on legacy technology systems.
That warning is especially significant for financial institutions because old infrastructure, complex software dependencies and high-value targets make banks attractive to attackers.
Bailey warned Mythos could reshape cyber risk
Bailey had already sounded the alarm before the reported briefing plan.
Reuters quoted him as saying in April that Anthropic may have found a way to “crack the whole cyber risk world open,” and that regulators needed to understand whether the new product could identify vulnerabilities that may be exploited for cyberattack purposes.
That concern helps explain why Anthropic is being asked to speak directly with finance ministries and central banks.
If Mythos can reveal weaknesses faster than traditional security teams can patch them, regulators may need to consider whether banks have enough time, resources and coordination to respond before attackers exploit the same classes of flaws.
Access remains limited as risks grow
Mythos remains tightly controlled partly because of security concerns, including a White House request that limited its distribution.
The access question is becoming a central issue. If only a small group of large banks and technology companies can use defensive AI tools like Mythos, smaller institutions may fall behind in identifying and fixing vulnerabilities.
The International Monetary Fund has warned such vulnerabilities could become macro-financial risks, especially for developing countries with weaker cyber defenses.
AI cyber tools become a regulatory problem
The briefing signals how quickly AI safety concerns are moving into financial regulation. The issue is no longer only whether models can write code or answer questions.
Regulators now have to assess whether AI systems that expose software flaws could alter the balance between defenders and attackers across the global banking system.
For Anthropic, the FSB briefing could help show that Mythos is being handled as a controlled defensive tool.
For regulators, it is a chance to understand whether the financial sector is prepared for a world where AI can find cyber weaknesses faster than institutions can fix them.
