US: New executive order asks AI firms to voluntarily run cybersecurity checks before releasing models

On June 2, 2026, U.S. President Donald Trump signed an executive order titled "Promoting Advanced Artificial Intelligence Innovation and Security", directing federal agencies to build a new oversight framework for safely deploying frontier AI models.

Key points of the order

• Establishes a process for AI developers to voluntarily give the U.S. government early access to new models — up to 30 days before release to other trusted partners.
• Directs federal agencies to build benchmarks to assess the cybersecurity capabilities of AI models.
• Creates an "AI cybersecurity clearinghouse" to review and share information about vulnerabilities.
• Strengthens cybersecurity defenses for the government's own systems.

Voluntary, not mandatory

An important point to note: this is a voluntary mechanism, relying on cooperation from major AI firms such as OpenAI, Anthropic and Google — the order explicitly states the government may not create licensing requirements or mandatory pre-clearance for new AI models. It is an offer to collaborate, not a legally binding regulation.

Why the shift in stance?

Notably, on his first day in office in 2025, Mr. Trump repealed the Biden administration's requirement that AI firms submit safety-test reports to the government, arguing it was too burdensome and stifled innovation. Returning to propose an oversight mechanism — even a voluntary one — shows that concern over cybersecurity risks from increasingly powerful AI models has forced the White House to reconsider the government's role in evaluating this technology.