Wall Street CEOs reportedly “summoned” to DC by Scott Bessent and Jay Powell to discuss AI cyber risks after Anthropic’s warning
Top officials are worried about left-tail cybersecurity risks from new AI tools, and are making sure the most important American bankers are taking the threat seriously.
The most powerful Americans in finance held an “urgent meeting” this week to discuss cybersecurity risks linked to new, powerful AI models — in particular, Anthropic’s Mythos.
Bloomberg reports that US Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell “summoned” the leaders of the biggest US banks — all of which are considered systemically important — to DC on Tuesday “to make sure banks are aware of possible future risks” and ensure that they “are taking precautions to defend their systems,” citing people familiar with the matter.
On Wednesday, Anthropic announced that it was releasing a version of its Mythos model to a select group of companies in an initiative called “Project Glasswing.” The hope is that these firms will get a head start on shoring up their defenses before malicious actors have the chance to strike with these new tools.
“AI models have reached a level of coding capability where they can surpass all but the most skilled humans at finding and exploiting software vulnerabilities,” warned Anthropic, which said it’s “found thousands of high-severity vulnerabilities, including some in every major operating system and web browser.”
Separately, OpenAI is also reportedly concerned that an upcoming cybersecurity tool of its own is too dangerous to be released publicly, and has similarly allowed a small group of its partners to test it out.
The CEOs of Citi, Morgan Stanley, Bank of America, Wells Fargo, and Goldman Sachs were said to be in attendance for this meeting at the Treasury Department. Jamie Dimon, CEO of JPMorgan (whose bank is a part of Project Glasswing), couldn’t make it.
Throughout 2026, we’ve discussed how the AI theme has become much more zero-sum in the stock market. For instance, AI demand has been helping memory and optics stocks, but fears of disruption have pushed software stocks down to multiyear lows.
The high degree of attention being paid to AI-fueled cybersecurity risks by top officials, with the same being demanded from their private sector counterparts, suggests that a similar lens may be appropriate to judge AI’s impact on the economy. That is, the potential for productivity benefits may need to be balanced against left-tail risks, particularly as agents are scaled and empowered to execute increasing workloads across companies.
That little sigh of relief you hear over in the corner is the private credit industry, grateful that these cybersecurity concerns mean there will be one less question asked about their own travails during banks’ quarterly conference calls when earnings season kicks off next week.
