Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell convened an emergency meeting of the chief executives of America's largest banks this week to discuss cybersecurity risks posed by Anthropic's newly released artificial intelligence model, Claude Mythos Preview.
The session, held at Treasury headquarters in Washington on Tuesday, was called at short notice while the bank chiefs were already in the capital for a Financial Services Forum board meeting, according to people familiar with the discussions.
Jane Fraser of Citigroup, Brian Moynihan of Bank of America, Ted Pick of Morgan Stanley, Charlie Scharf of Wells Fargo, and David Solomon of Goldman Sachs all attended, according to Bloomberg, which first reported the meeting; JPMorgan chief executive Jamie Dimon was invited but could not attend.
Officials used the meeting to ensure banks were aware of the potential risks from Mythos and were taking steps to defend their systems, sources said.
Anthropic launched Claude Mythos Preview on 7 April in a deliberately limited capacity, citing concern that the model's ability to identify and exploit software vulnerabilities could be misused if released widely.
The company said it would extend access to around 40 organisations that build or deploy critical software infrastructure, alongside a group of major technology and financial partners, including JPMorgan Chase, Amazon, Apple, Google, and Nvidia.
The initiative, called Project Glasswing, is described by Anthropic as an effort to mobilise the model's capabilities for defensive purposes before similar technology becomes available to malicious actors.
Anthropic committed up to $100 million in Mythos usage credits to partners in the programme.
A Treasury spokesman said the department is pushing financial institutions to prepare for new security threats, adding that further coordination meetings are planned across regulators and institutions.
Related reading
- OpenAI's Sam Altman warns AI bioweapon and cyberattack risks are 'no longer a theoretical thing'
- Anthropic secures multi-gigawatt Google and Broadcom chip deal as annual revenue run rate hits $30bn
- OpenAI warns macOS users to update apps after supply chain security breach
The meeting takes place against a backdrop of escalating tensions between Anthropic and the federal government: a US federal appeals court this week rejected the company's attempt to block a Pentagon designation of Anthropic as a supply chain risk to national security, a label applied after the company refused to permit its technology to be used for autonomous weapons.
Anthropic has said it remains committed to working with the government to ensure national security is prioritised as it develops and deploys new AI models.
The recap
- Regulators summoned big bank CEOs to Washington this week.
- Anthropic's Mythos will be offered to a few dozen companies.
- Officials flagged AI-driven cyber risks; meeting may prompt regulation.
