President Donald Trump has postponed signing an executive order designed to establish government security evaluations for artificial intelligence models before their public release, expressing reservations about the order’s language and potential impacts on US technological leadership.
- Trump delays AI security order citing language concerns
- Proposed order demands advanced model sharing 14-90 days before launch
- Goal is to safeguard against AI security vulnerabilities without hindering US leadership
What happened
President Donald Trump decided to delay signing an executive order that would have mandated pre-release security reviews of artificial intelligence models by government agencies. The decision came after Trump expressed dissatisfaction with certain elements in the language of the order, which he felt could act as a blocker to innovation and leadership in AI technology.
Reports indicate that another factor influencing the delay was the absence of sufficient participation from top technology CEOs who could not make it to Washington on short notice for the planned signing ceremony, which traditionally includes a photo opportunity to highlight collaboration between government and industry.
Why it matters
The executive order was intended to empower the Office of the National Cyber Director and other federal bodies to implement a formal process for evaluating potential security risks in new AI models before their release. This effort comes as concerns mount over the ability of AI models such as Anthropic’s Mythos and OpenAI’s GPT-5.5 Cyber to rapidly expose and exploit vulnerabilities in digital infrastructure.
A key contentious point in the proposed directive was the requirement for AI developers to share details of advanced models with the government 14 to 90 days prior to public launch. Trump voiced concern that this requirement might stifle innovation by imposing excessive pre-launch government oversight, which could slow the US’s competitive edge in the global AI race, particularly against rivals like China.
What to watch next
Stakeholders will be closely monitoring any revisions to the executive order’s language that address government and industry concerns while still safeguarding national security and user safety. How the administration balances the need for proactive AI risk management with incentives for innovation will be crucial to the order’s future viability.
Additionally, the Biden administration’s approach to engaging CEOs and tech leaders in shaping the policy framework will be important. Future announcements may include new timelines for signing the order, adjusted pre-release sharing requirements, or alternative mechanisms for collaboration between the government and AI companies.