Will Trump control new AI models ?

Picture Credit: WIRED

The White House is preparing to take one of its most significant steps yet in regulating artificial intelligence and it may arrive as soon as this week.

According to sources familiar with the matter, the Trump administration is drafting an executive order that would ask AI companies to voluntarily share their most advanced models with the federal government before releasing them to the public. The proposal centers on a pre-launch review window, with the government pushing for 90 days and major AI firms preferring something closer to 14 days.

The urgency is real. Cybersecurity experts have long warned that cutting-edge AI can dramatically amplify the scale and speed of cyberattacks. That concern moved from theoretical to tangible when Anthropic unveiled its Mythos model earlier this year, one the company itself acknowledges can exploit security vulnerabilities at an unprecedented pace. Rather than release it publicly, Anthropic is offering tightly controlled access through a private consortium. OpenAI has taken a similar approach, granting select businesses and governments early access to its latest models to help shore up cyber defenses.

The draft executive order is said to cover two key areas: a cybersecurity "clearinghouse" which is a joint effort between the Treasury Department, other agencies, and AI companies to identify and patch vulnerabilities in unreleased models, and a framework defining which frontier AI systems would fall under the voluntary review process.

The administration's shift is notable. For much of its tenure, it has favored a hands off approach to AI oversight. But as the technology grows more powerful and the risks more concrete, even Washington is moving toward a closer look before the rest of the world gets one.