Trump delays AI security executive order: ‘I don’t want to get in the way of that leading’
Original reporting by TechCrunch
President Donald Trump has postponed the signing of a pivotal executive order aimed at regulating the release of advanced artificial intelligence models, citing dissatisfaction with its current language. The order, which would mandate government evaluation of AI for security risks prior to public deployment, has been a key policy focus amidst growing concerns over powerful, potentially exploitable models like Anthropic's Mythos and OpenAI's GPT-5.5 Cyber.
Trump expressed reservations that the proposed directive could "get in the way" of American leadership in AI, emphasizing the nation's competitive edge against countries like China. "I didn't like certain aspects of it," he told reporters, suggesting the language "could have been a blocker" for innovation. Meanwhile, reports indicate an unofficial reason for the delay: a logistical challenge in gathering a sufficient number of prominent tech CEOs for the customary White House photo opportunity.
The Order's Intent
The anticipated executive order would have empowered the Office of the National Cyber Director, among other agencies, to establish a framework for pre-release security assessments. A central point of contention within the order’s draft language was a requirement for AI developers to share their advanced models with the government between 14 and 90 days ahead of launch. This provision, intended to mitigate national security risks, appears to be at the heart of the administration's current hesitation, balancing oversight with the rapid pace of technological advancement.
The delay in President Trump’s executive order on AI evaluation signals more than just a disagreement over specific language; it underscores the profound challenges inherent in governing rapidly evolving artificial intelligence. This pause leaves a critical regulatory vacuum, prolonging uncertainty for developers while allowing potential national security vulnerabilities, exemplified by models like Anthropic’s Mythos and OpenAI’s GPT-5.5 Cyber, to remain unaddressed by a formal framework. The administration's public concern about "blockers" highlights a fundamental tension between fostering unfettered innovation and implementing necessary safeguards to mitigate systemic risks.
The Regulatory Divide
This situation necessitates a continued, informal balancing act, with companies potentially navigating a patchwork of voluntary safety measures rather than a clear federal mandate. The proposed requirement for pre-release model sharing, a key sticking point, touches upon the core debate surrounding intellectual property, competitive advantage, and government oversight in the tech sector. Furthermore, the delay’s unofficial link to a missed photo opportunity suggests that the rollout of significant AI policy remains susceptible to political optics, potentially obscuring the pressing technical and ethical considerations at play. As the global race for AI supremacy intensifies, every regulatory decision is scrutinized for its impact on national leadership. The ultimate shape of U.S. AI governance will profoundly influence not only the security and ethical development of these powerful technologies but also America's position on the world stage for decades to come.