White House Signals More Hands

White House Signals More Hands-On Approach to Oversight of AI Systems

From hands-off to hands-on. The Trump administration increasingly is taking a direct role in overseeing the release of the most powerful artificial intelligence (AI) models, marking a sharp turnaround from its previous policy of removing regulations that might hamper U.S. development of the technology.

Last week, the White House lifted the export restrictions it had imposed on Anthropic’s Mythos 5 and Fable 5 models, allowing the company to restore broader access to the powerful AI systems in the U.S. and overseas. The reversal came after Anthropic agreed to include guardrails meant to limit the chances that Fable 5’s cyber capabilities would be abused.

At nearly the same time, the administration reached an agreement with OpenAI to limit access to the ChatGPT-maker’s latest model, GPT 5.6, to a handful of U.S. companies and organizations approved to by the White House. According to Politico, OpenAI initially had not planned to restrict access to the general-use model, but hit pause on the rollout at the White House’s request and in consultation with the Office of Science and Technology Policy and the Office of the National Cyber Director.

The latest White House moves offer a stark contrast with the administration’s initial stance regarding AI development. Shortly after taking office, Trump rescinded his predecessor’s executive order establishing federal standards for AI safety and security, Americans’ privacy, advancing equity and civil rights, protecting consumers and workers, promoting innovation and competition, and advancing American AI leadership. In its place, the Trump administration issued a National AI Legislative Framework calling for an across the board rollback of federal regulations it considered unduly burdensome on AI developers and development and sought to preempt state laws regulating the technology.

The release of Mythos and Fable, however, with their capability to identify and exploit cyber vulnerabilities, triggered a reevaluation within the White House of its hands-off approach.

According to The National News Desk, supporters of the new approach say stronger oversight is necessary to keep pace with rapidly evolving capabilities and to prevent potential national security threats before they emerge, although critics remain skeptical of direct government direction of AI developers’ release plans.

Either way, both AI companies and the federal government see a need to renegotiate the terms of engagement. Last week, OpenAI floated a proposal to hand the government a 5% stake in the company in a bid to defuse growing tensions with the White House. The stake would be worth $42.6 billion based on OpenAI’s current $852 billion valuations. OpenAI CEO Sam Altman had previously discussed the government taking similar size stakes in each of the leading AI companies, including Anthropic, Google and Meta.

In a June executive order, meanwhile, the White House announced a policy “to promote AI innovation and security by working collaboratively with the private sector to modernize government and private sector information systems and harden them against external threats.” As part of that new policy, the order directed the Departments of Homeland Security and Defense, along with the Committee on National Security Systems, to take “appropriate and expeditious action” consistent with the purpose of the order.

As regulators and industry leaders each try to navigate a changing landscape, reordering the balance among innovation, security and oversight is rapidly becoming one of the defining policy questions of the AI era.