04.10.26

Banks Warned About Anthropic’s New AI Technology

by: Rob Copeland; Colby Smith

Senior U.S. officials have warned bank executives about potential cybersecurity risks from a powerful new AI model developed by Anthropic. At a meeting, officials including Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell cautioned bank bosses that the system could expose vulnerabilities in financial institutions if integrated into internal networks. The model, called Claude Mythos Preview, is designed to identify software security flaws that humans might miss, but officials warned that such capabilities could also be exploited by hackers if the underlying insights are misused. Anthropic has restricted access to the model through a limited partnership program called “Project Glasswing,” which includes select firms such as JPMorgan Chase.

Read the full article on The New York Times