Federal Reserve Chair Jerome Powell and Treasury Secretary Scott Bessent convened an emergency closed-door meeting with the chief executives of the United States' largest banks on Tuesday, 8 April, to discuss cybersecurity risks posed by Anthropic's latest artificial intelligence model, Mythos. The hastily arranged session, held at the Treasury Department's Washington headquarters, was called to ensure major financial institutions were aware of potential threats and were taking precautions to protect their systems, according to multiple sources familiar with the matter who spoke on condition of anonymity.
The bank heads were already in Washington for a Financial Services Forum board meeting when the special gathering was convened to discuss Mythos. Those in attendance included Citigroup chief executive Jane Fraser, Morgan Stanley chief executive Ted Pick, Bank of America chief executive Brian Moynihan, Wells Fargo chief executive Charlie Scharf and Goldman Sachs chief executive David Solomon. JPMorgan Chase chief executive Jamie Dimon was invited but was unable to attend.
What is Mythos?
Anthropic has described Mythos as a frontier model that can outperform "all but the most skilled humans at finding and exploiting software vulnerabilities," claiming it has already identified thousands of software flaws previously unknown to their developers, including some that were decades old inside companies widely considered to be security strongholds. Anthropic said it will not widely release Mythos because of its advanced capabilities, which have uncovered vulnerabilities in major operating systems and web browsers.
Anthropic released a report titled "Assessing Claude Mythos Preview's cybersecurity capabilities," noting how the model was able to find vulnerabilities including a 27-year-old flaw in OpenBSD, an operating system with a strong reputation for security. The company released the model in limited capacity, making it available initially to a select group of technology and financial firms under an initiative called Project Glasswing. Partners in the project include Amazon, Apple and Nvidia, with the aim of using Mythos to strengthen cybersecurity defences. The idea is that these firms will get a head start on strengthening their defences before the model becomes more widely available or similar tools emerge from other developers.
An Anthropic official told CNBC the company is in "ongoing discussions" with the US government, including the Cybersecurity and Infrastructure Security Agency and the National Institute of Standards and Technology's AI Safety Institute, regarding the capabilities of Claude Mythos Preview.
A broader confrontation
The Trump administration's engagement over the Mythos model comes as Anthropic is locked in a dispute with the Department of Defense, which recently labelled the company a supply chain risk to national security. President Donald Trump and Defence Secretary Pete Hegseth have publicly criticised the company for insisting on limits to the use of its AI in warfare. A federal appeals court this week denied Anthropic's request to temporarily block the blacklisting, though a separate ruling allows the company to continue working with other government agencies while the legal challenge proceeds.
The surprise meeting between the bank chiefs and the two most powerful federal monetary regulators was seen as a signal that the advanced capabilities of AI are a top concern in the Trump administration and could threaten the foundation of the US financial system. In his annual letter to shareholders published this week, JPMorgan's Dimon warned that cybersecurity "remains one of our biggest risks" and that "AI will almost surely make this risk worse."
Sources: CNBC, Bloomberg, CBS News, Fortune, Reuters