Treasury Secretary Scott Bessent cautioned that rapid advances in artificial intelligence are exposing new vulnerabilities that could enable large-scale bank account hacks. He said the U.S. government is working with AI companies and regulators after a recent leap in one large language model's capabilities. Bessent and Federal Reserve Chair Jerome Powell held a quick meeting with Wall Street bank leaders to discuss risks tied to Anthropic's Claude Mythos Preview, which reportedly found more than 2,000 previously unknown software flaws during seven weeks of testing. Officials emphasized the need to balance innovation with safety and to strengthen cooperation among regulators, banks and AI firms.
Treasury Secretary Scott Bessent Warns of AI-Driven Bank-Account Hacks After New Model Finds 2,000+ Flaws
Treasury Secretary Scott Bessent speaks at the 56th World Economic Forum Annual Meeting in Davos, Switzerland, on Jan. 20.(Getty Images)
Treasury Secretary Scott Bessent warned Sunday that rapidly advancing artificial intelligence tools are exposing fresh vulnerabilities in financial systems that could enable large-scale bank account hacks. He said the U.S. government is coordinating closely with AI companies and financial regulators to address emerging threats.
"The U.S. government has gotten involved. The AI companies are working with us," Bessent said on Sunday Morning Futures, noting that recent jumps in AI capability can create unexpected risks.
The warning followed a quickly convened meeting in which Bessent and Federal Reserve Chair Jerome Powell gathered senior Wall Street bank leaders to discuss cybersecurity concerns tied to Anthropic's new model, Claude Mythos Preview. Reports tied to Anthropic's testing indicate Mythos identified more than 2,000 previously unknown software vulnerabilities during seven weeks of evaluation — a finding that underscores the scale of potential exposure.
Purpose of the Meeting
Bessent characterized the flash meeting as "less dramatic" than initial coverage suggested but said its purpose was practical: to ensure major financial institutions are preparing for emergent, AI-driven cyber threats and to coordinate immediate mitigation steps.
An image showcasing Anthropic AI on a laptop computer screen.
Balancing Innovation and Safety
Echoing other administration officials, Bessent stressed the strategic importance of keeping the United States at the forefront of AI development, particularly relative to China. "There is a very important calculus here between innovation and safety," he said, adding that the government's role is to preserve safety while allowing AI companies to innovate.
What Experts Say
Cybersecurity specialists have warned that highly capable AI systems can accelerate the discovery and exploitation of software flaws, enabling attackers to operate at greater speed and scale. In response, regulators, banks and AI firms are increasing cooperation to harden defenses, share threat intelligence, and improve incident response capabilities.
Key actions underway include enhanced information sharing between government and industry, stress-testing of critical banking systems for AI-discovered vulnerabilities, and joint planning to mitigate any newly revealed attack vectors.
Fox News' Robert McGreevy contributed to this report.
Help us improve.
