Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell have summoned the leaders of the nation's largest financial institutions to an urgent, short-notice meeting to address concerns about potential cyber risks posed by Anthropic's unreleased Mythos AI model and similar advanced models. The gathering at the Treasury's headquarters in Washington marks another sign that federal regulators now view a new class of highly capable AI systems as a top-tier systemic risk for the financial industry.
According to Bloomberg, officials convened the meeting to ensure that major banks are aware of the risks associated with Anthropic's latest model and are taking defensive precautions. The concerns center on Mythos, which Anthropic says is capable of identifying and exploiting vulnerabilities across every major operating system and web browser when directed by a user. In its own technical write-up, Anthropic said the model also proved capable of autonomously discovering many vulnerabilities during testing, including exploit chains that were previously the sort of work associated with expert offensive security researchers.
Anthropic's security team said Mythos was able to compromise a web browser so that a malicious site could read sensitive data from another website, such as a victim's bank. The company also described the model as capable of autonomously identifying zero-day vulnerabilities and, in some cases, turning them into working exploits without human intervention after an initial prompt. Anthropic said it found the model's capabilities had emerged as a downstream effect of broader gains in code generation, reasoning, and autonomy, rather than from explicit training for offensive cyber tasks.
Powell's participation underscored that the matter is being treated as a systemic issue, not a narrow industry concern. Many of the executives were already in Washington for a meeting of the Financial Services Forum, an advocacy group made up of the country's largest lenders. Bloomberg reports that CEOs from Goldman Sachs, Morgan Stanley, Citigroup, Bank of America, and Wells Fargo attended, while JPMorgan's Jamie Dimon was unable to attend.
Anthropic has responded by restricting Mythos to a small group of major technology and finance firms through its Project Glasswing initiative. The group includes Apple, Amazon, and JPMorgan, and is intended to help secure critical systems before models with similar capabilities become more broadly available. Anthropic said it does not plan to make Mythos generally available at this time.
The move also puts fresh attention on the security implications of increasingly autonomous software tools. Apple has recently embraced AI-assisted development, including with Xcode 26.3, while continuing to issue urgent fixes for real-world threats such as its recent iOS 18.7.7 security update. Anthropic's Mythos findings highlight how the same advances making AI more useful for defenders and developers could also sharply increase the speed and scale of offensive cyber activity.
Other regulators are moving as well. Bloomberg says the Bank of Canada met Friday with major financial firms to discuss the cybersecurity implications of Mythos, and the Bank of England is preparing its own meeting with banking and insurance executives.
![13-inch MacBook Air M5 Drops to $949 in New All-Time Low [Deal]](/images/news/100466/100466/100466-160.jpg "13-inch MacBook Air M5 Drops to $949 in New All-Time Low [Deal]")
