The Trump administration is considering requiring federal oversight of artificial intelligence models before their public release, a significant reversal from its earlier hands-off approach, The New York Times reported Monday, citing U.S. officials and individuals familiar with the matter. The White House is discussing an executive order that would create a working group of tech executives and government officials to evaluate review procedures for advanced AI systems.

Senior administration officials recently briefed executives from Anthropic, Google, and OpenAI on the potential framework during meetings last week, the report said. The shift comes after the emergence of Anthropic's powerful new model, Mythos, which the company has declined to release due to concerns over its ability to exploit software vulnerabilities and trigger major cybersecurity threats.

The move marks a departure from President Trump’s earlier position as a champion of unfettered AI development; he previously rolled back Biden-era safety testing requirements and criticized regulation as harmful to U.S. competitiveness with China. The Biden administration had mandated AI developers to share safety test results with the government—a policy Trump rescinded upon taking office.

The reconsideration follows rising bipartisan concern in Congress and public anxiety over AI’s effects on jobs, education, and national security. Internal administration dynamics have also shifted, with former AI czar David Sacks departing in March; White House Chief of Staff Susie Wiles and Treasury Secretary Scott Bessent are now taking a more active role in AI policy formulation.

A White House official told The Times that talk of an executive order remained speculative and that President Trump would announce any policy change personally. The administration continues to navigate a legal dispute between Anthropic and the Pentagon over a $200 million military AI contract as it shapes its approach.