Federal Reserve Chairman Kevin Warsh testifies during the Senate Banking, Housing and Urban Affairs Committee hearing entitled “The Semiannual Monetary Policy Report to Congress” on Wednesday, July 15, 2026, in the Dirksen Building.
Tom Williams | Cq-roll Call, Inc. | Getty Images
In April, the Federal Reserve and Treasury Department convened an extraordinary meeting with the CEOs of the nation’s leading banks. Officials sounded the alarm about an advanced new artificial intelligence model that could pose an unprecedented cybersecurity threat to the country’s leading financial institutions.
Anthropic, the company behind the Claude Mythos Preview AI model, said the offering excels at identifying vulnerabilities and vulnerabilities in software. The company shared it with a select group of banks and other institutions as part of a cybersecurity initiative called Project Glasswing.
For at least three months, the Fed itself had no access to Mythos, leaving arguably the most systemically important global financial institution vulnerable even as other institutions began to address their weaknesses.
The Fed was still trying to gain access to Mythos as of July 15th. It is not clear whether the central bank has since gained access to the model.
Anthropic did not immediately respond to a request for comment. The Fed declined to comment for this article.
Table of Contents
Read more about CNBC’s politics coverage
The April meeting was led by former Fed Chairman Jerome Powell, who along with Treasury Secretary Scott Bessent convened bank CEOs, CNBC previously reported. But in little-noticed testimony to Congress last week, Powell’s successor, Chairman Kevin Warsh, told the Senate that he was still working to secure access to Mythos and other cutting-edge AI models.
“We are not the decision makers about who has access, but I have not been afraid to share my views on the vulnerabilities with agencies across the government and have asked for access to a whole range of these new artificial intelligence models, not just for the Federal Reserve, but for other institutions so that they can protect themselves,” Warsh told Sen. Jack Reed, D-R.I., in response to questions about Mythos.
Warsh made it clear that the Fed also needs to use other models.
“But I don’t want to just isolate Mythos,” Warsh said. “As these new models become more widespread, our banking system and, frankly, the Federal Reserve must do everything we can to address any vulnerabilities we have.”
Warsh has welcomed the introduction of AI in his short time at the Fed, calling it a transformative technology.
It is not clear whether work to fix vulnerabilities can begin without access to Mythos.
Anthropic introduced Claude Mythos Preview and Project Glasswing in early April.
The company said about 50 organizations had access to the model at the time, but only named a handful, including the bank JPMorgan Chaseand tech titans like Amazon, Apple And Google. Anthropic said it has also been in “ongoing discussions with U.S. government officials” about the model, including the Cybersecurity and Infrastructure Security Agency and the Center for AI Standards and Innovation.
Anthropic expanded access to Project Glasswing in June, adding more than 150 organizations in 15 countries to the initiative. Daniel Newman, CEO of research firm Futurum Group, said he was surprised to hear the Fed wasn’t included.
“You would think that the financial institution that sets all the policy for the rest of the financial institutions would be at the center and at least have a chance to evaluate the new technology,” he told CNBC on Tuesday.
The introduction of Mythos has caused both intrigue and confusion in recent months, particularly as Anthropic’s strained relationship with the Trump administration led to complications.
In June, Anthropic said it had to disable access to Mythos 5, an updated version of the model, and Fable 5, a version of Mythos it released more broadly, to comply with a federal government export control directive that cited “national security agencies.”
According to a letter seen by CNBC, Commerce Secretary Howard Lutnick gave the company permission to restore access to Mythos to a select group of “trusted partners.” Export controls were later completely lifted.
Since President Donald Trump signed an AI executive order in June, the Trump administration has taken a much more active role in AI regulation. But questions remain about who is driving AI policy behind the scenes, and the wrangling at the Fed — which Trump tapped Warsh to lead — is another sign of chaos.
Chris Fall, head of the Center for AI Standards and Innovation, stepped down from his role as director, just three months after being tapped for the job by the Trump administration, CNBC confirmed Monday. Venture capitalist David Sacks previously held the position of White House AI and crypto czar but resigned from the position in March.
The pressure is building.
Chinese open-weight models are gaining traction over leading offerings from U.S. companies such as OpenAI and Anthropic, raising concerns among tech executives and government officials about the durability of the U.S. lead in the AI competition. Moonshot AI, a Chinese startup, released a model called Kimi K3 earlier this month that outperforms these companies in some industry benchmarks.
Sacks said in a post on X on Friday that Kimi K3’s performance was “worrying” and that “America is getting itself into trouble.”
“This is how you lose the AI race,” he wrote. “The rest of the world won’t play by our rules if we mess up.”
Futurum’s Newman said the Fed will “certainly catch up” since it doesn’t have access to other agencies’ cutting-edge models.
“Every day, every week, whether it’s innovation in China or in the U.S., technology leaders at large institutions face a constant flood of net innovation,” he said.
Choose CNBC as your preferred source on Google and never miss a moment from the most trusted name in business news.
https://www.cnbc.com/2026/07/21/fed-mythos-ai-cybersecurity-banks-project-glasswing.html
