Bessent and Powell caution Wall Street executives about the dangers of Anthropic AI, according to reports

Emergency Meeting on AI Cybersecurity Risks

Treasury Secretary Scott Bessent and Federal Reserve Chairman Jerome Powell held an urgent meeting on Tuesday with major Wall Street bank leaders in Washington, D.C. This gathering was prompted by growing cybersecurity threats associated with Anthropic, a prominent AI company, according to a Bloomberg report released Thursday night.

During this impromptu meeting at the Treasury’s headquarters, Bessent and Powell aimed to prepare the banks for potential risks that could arise from Anthropic’s new model, Claude Mythos Preview, which experts believe represents a significant technological advancement.

The banks involved are considered “structurally important” to the global financial infrastructure, with chief executives from firms like Goldman Sachs, Citigroup, Morgan Stanley, Bank of America, and Wells Fargo in attendance.

Notably, Brian Moynihan, the CEO of Bank of America, was present, although responses from Goldman Sachs and Wells Fargo were not forthcoming. Citigroup and Morgan Stanley had no comments regarding the meeting.

Military Technology and AI

Interestingly, Jamie Dimon, CEO of JPMorgan Chase, was also asked to attend but couldn’t make it. Sources say that JPMorgan is part of Anthropic’s “Project Glasswing,” which aims to utilize Mythos to defend against similar AI models in the future. JPMorgan, too, did not comment on the matter.

Mythos has generated considerable online interest because Anthropic claims it can autonomously find and take advantage of software vulnerabilities. They label Mythos as a “frontier model,” asserting that it surpasses the abilities of most skilled humans for identifying software flaws, including some that have been hidden for decades.

According to a blog post from Anthropic, this technology could lead to more frequent and harmful cyberattacks, which might bolster adversaries against the U.S. and its allies. Addressing these security concerns, they note, is crucial for democracies.

Anthropic’s Struggles with the Pentagon

A source close to Anthropic indicated that the company has briefed U.S. government officials about the inherent risks of Mythos, although they declined to specify which agencies were involved.

Previously, Anthropic was a core partner for the U.S. military and signed a $200 million contract with the Department of Defense in July 2025. However, that partnership soured, leading to their split in February after disagreements over the use of their technology for autonomous weaponry and domestic surveillance. Following an ultimatum, the Army Secretary designated Anthropic as a supply chain risk, banning its use among federal contractors.

Efforts by Anthropic to contest this designation were recently rejected by a federal appeals court. On Tuesday, a representative referred to a statement from Acting Attorney General Todd Blanche, who praised Wednesday’s ruling on the matter.

In a statement, Blanche highlighted the significance of the ruling for military readiness, asserting that if Anthropic’s technology was to be integrated into sensitive governmental systems, full access must rest with the military. Control, he emphasized, should ultimately reside with military leadership.