The U.S. Department of the Treasury seeks access to Mythos models under Anthropic to identify system vulnerabilities.

Deep Tide TechFlow news, April 14, according to an informed source, the U.S. Treasury Department’s technical team is seeking to access Anthropic’s Mythos artificial intelligence model in order to begin looking for system vulnerabilities.

The source said the goal of U.S. Chief Information Officer Sam Cocos of the Treasury Department is to obtain access to the model as soon as possible (at the earliest, possibly this week). Anthropic has long allowed only a limited number of institutions to access the model. The source said that Cocos introduced the technology to the Treasury’s cybersecurity team last week and instructed them to prepare for the ultimate threats that powerful AI systems may pose.

U.S. Treasury Secretary Bessent and Federal Reserve Chair Powell convened an emergency meeting with Wall Street leaders last week, fearing that the Mythos model would usher in an era with higher cyber risks. Anthropic has issued a warning that if companies do not test their own systems and establish defenses before broad deployment, the model may have the capability to support cyberattacks.(Jin10)