Trump administration weighs Pentagon safety evaluations for AI models used by governments

The administration of US President Donald Trump is weighing a proposal that would mandate the Pentagon to conduct safety evaluations of artificial intelligence (AI) models used by federal, state, and local governments, Axios reported on Tuesday. The White House's Office of the National Cyber Director (ONCD) held two meetings last week with technology and cybersecurity firms, as well as tech industry groups, to address wider security issues linked to advanced AI systems.

Executive order exploration

Washington is also exploring an executive order that could task multiple agencies with conducting safety assessments of new AI models, according to the report. The report noted that efforts for the framework were underway before Anthropic's Mythos was released, triggering fresh cybersecurity concerns.

White House response

A White House official said that any policy announcement "will come directly from the president" and that discussion about "potential executive orders is speculation." Since Anthropic unveiled Mythos, the White House has been trying to assess its potential for hacking and determine what role the administration should play in managing the most significant national security risks.

Agency access

At the same time, the White House has been considering executive measures that would enable federal agencies to bypass the current restriction on using Anthropic tools, allowing them to adopt Mythos. The AI model has raised concerns due to its ability to identify and exploit vulnerabilities in major operating systems and web browsers.