Trump signs AI safety EO June 2 — frontier labs asked to voluntarily submit most powerful models for 30-day cyber review before release
President Trump signed an executive order on June 2, 2026 asking AI companies to voluntarily submit their most powerful models for government testing up to 30 days before public release. The order marks a shift from the administration's hands-off posture and was cut from an earlier 90-day-review draft that the White House had paused in May over concerns about slowing American AI companies against Chinese competition.
The 30-day window is the substantive number. The version of the order that the White House prepared to sign in early May would have given the government up to 90 days to review advanced models before release; the final order cuts that to 30. The reduction reflects the political tradeoff Trump described publicly — concern that the longer window would stifle the American lead during the China frontier-model race. Even at 30 days, the review is opt-in: labs choose whether to submit, and the order creates no binding pre-market approval gate.
The order's operational scope is cybersecurity-first. Federal agencies are directed to develop benchmarks for AI cyber capabilities, stand up an "AI cybersecurity clearinghouse" to centralize vulnerability information across labs, and harden government security defenses against AI-enabled attacks. The framing is national-security testing rather than the broader safety evaluation that the original Biden-era safeguards (repealed in January 2025) attempted. CAISI — housed inside NIST since June 2025 — is the implied executor; it already has voluntary MOUs with Anthropic, OpenAI, and the May 2026 additions of Google DeepMind, Microsoft, and xAI.
The voluntary frame keeps the new order aligned with the broader US deregulatory direction, including EO 14365 and the AI Litigation Task Force challenging state AI laws. The combined posture: no binding federal pre-market rules, the federal preference preempts stricter state rules, and the safety review the federal government does run is a 30-day voluntary look at cyber capabilities.
NPR — Trump's new AI safety order seeks voluntary review of new models → · The White House — Promoting Advanced Artificial Intelligence Innovation and Security → · Roll Call — Executive order sets voluntary cyber reviews for advanced AI →