On Monday the New York Times revealed the White House is weighing mandatory pre‑release reviews of AI models, a sharp turn from the Trump administration’s year‑long push to dismantle Biden’s AI safety order in early 2024, to the public. The reversal follows the removal of David Sacks, the venture‑capitalist‑turned‑AI‑and‑crypto czar, whose one‑year stint had tried to block state AI laws and centralize policy in the White House.

The leak of Anthropic Mythos, a model that could autodetect cybersecurity flaws, alarmed national‑security officials. Treasury Secretary Scott Bessent and Chief of Staff Susie Wiles met Anthropic CEO Dario Amodei in April, signaling that the threat outweighed earlier political battles over “woke” AI. Analysts say once senior security figures engage, policy debates become harder to politicize for the United States as a priority.

On Tuesday the Commerce Department appointed the Center for AI Standards and Innovation (CAISI) to conduct pre‑deployment testing of frontier models, already securing agreements with xAI, Microsoft and Google DeepMind. By restoring authority to agencies like NIST, the administration aims to balance innovation with national‑security safeguards, marking a decisive shift from the deregulation agenda that Sacks championed across critical infrastructure sectors.