Related ToolsChatgptClaude

Trump Delays AI Security Review Order, Citing Risk to U.S. AI Lead

Editorial illustration for: Trump Delays AI Security Review Order, Citing Risk to U.S. AI Lead

President Trump declined to sign an executive order that would have required AI companies to submit their models for government security review before public release. He told reporters he was dissatisfied with the order's language and didn't want regulation to interfere with the U.S. maintaining its lead in AI development.

The order had been drafted to establish mandatory pre-release screening of large AI models for national security risks - effectively filling the gap left when Trump revoked Biden's 2023 AI executive order in January 2025. Biden's order had included provisions requiring AI developers to share safety test results with the government before deployment. Those requirements are now gone, and this week's delay means no replacement is forthcoming in the near term.

The Gap This Leaves

There is currently no federal requirement for AI companies to have their models reviewed before release. OpenAI, Anthropic, Google, and Meta can publish frontier models - including the most capable ones - without any mandatory government security check.

Researchers have flagged specific capabilities at the frontier as worth screening: assistance with designing biological agents, generating cyberweapons, and running large-scale influence operations. The major labs have voluntary frameworks addressing these - Anthropic publishes a Responsible Scaling Policy, OpenAI has a Preparedness Framework - but compliance is self-reported and no external body verifies the results.

Trump's reasoning tracks the argument AI industry lobbyists have pushed hard: mandatory review requirements slow American companies while Chinese AI developers face no equivalent domestic process. It's a real competitive dynamic. It also aligns exactly with what the industry wants, which is worth noting when evaluating how much weight to give it.

Congress has not passed comprehensive federal AI legislation. State-level attempts, including California's SB 1047 in 2024, have largely collapsed under industry pressure. As of today, the U.S. has no federal AI safety review requirement and no clear timeline for one.