Treasury Secretary Scott Bessent reportedly met with banking leaders this week to discuss the cyber risks posed by Anthropic's latest artificial intelligence model amid growing concern that rapid advances in AI could outpace safeguards designed to protect the U.S. financial system.
According to a report Friday by the Financial Times, Bessent convened executives from Bank of America, Citigroup, Goldman Sachs, Morgan Stanley, and Wells Fargo while they were in Washington for a banking industry meeting.
Federal Reserve Chair Jay Powell also attended. JPMorgan Chase CEO Jamie Dimon was invited but unable to participate.
The urgency behind the meeting centers on Anthropic's newly unveiled AI system, "Claude Mythos Preview," which experts say has demonstrated a strong ability to identify software vulnerabilities — capabilities that could be used defensively but also weaponized by bad actors.
Anthropic acknowledged that its model can surpass most human programmers in detecting exploitable weaknesses, reportedly uncovering thousands of severe vulnerabilities across major operating systems and web browsers, including some that had gone undetected for decades.
The Trump administration's concern reflects a broader warning from financial and national security leaders that artificial intelligence may dramatically lower the barrier to launching sophisticated cyberattacks.
Dimon, in his annual shareholder letter, reiterated that cyber threats remain one of the biggest risks to the banking system and cautioned that AI will likely "make this risk worse," requiring significant new investments in defense.
The New York Times reported this week that top tech companies have also been briefing administration officials on the risks tied to advanced AI systems like Mythos.
While such tools could strengthen cybersecurity by identifying weaknesses before hackers do, experts warn they could just as easily be exploited by hostile nations or criminal networks.
"What used to require the resources of major governments or large criminal enterprises could soon be within reach of smaller actors," one expert told the Times, raising alarms about the "democratization" of cyberattack capabilities.
Anthropic has attempted to limit those risks by restricting access to the model, providing it only to a select group of major companies, including Amazon, Apple, and Microsoft.
The company has also launched a collaborative initiative aimed at patching vulnerabilities before they can be exploited.
Still, recent data leaks tied to the startup have heightened concerns about whether even well-intentioned safeguards are sufficient.
For conservatives, the developments reinforce longstanding warnings about unchecked technological expansion and the national security vulnerabilities it may create — particularly as adversaries such as China, Russia, and Iran seek to exploit emerging tools.
The administration has increasingly framed AI as both an economic opportunity and a strategic threat, signaling that tighter oversight and stronger public-private coordination may be necessary to stay ahead.
Charlie McCarthy ✉
Charlie McCarthy, a writer/editor at Newsmax, has nearly 40 years of experience covering news, sports, and politics.
© 2026 Newsmax. All rights reserved.