White House Pressures OpenAI to Slow-Roll GPT 5.6 Release Over Safety Risks

The White House has pressured OpenAI to stagger the rollout of its latest model, GPT 5.6, restricting early access to select partners rather than a public launch, according to reporting from The Information.

At an internal meeting, CEO Sam Altman disclosed that the government would oversee access "customer by customer" during a preview phase, with a broader release potentially following "a couple of weeks later" if the trial proves successful.

The move represents a tangible shift in the Trump administration's AI stance. Though the White House initially framed itself as taking a hands-off approach, it has since advocated for federal review. Earlier this month, Trump signed an executive order directing certain AI firms to voluntarily submit new models for government testing before public deployment. The agencies involved in OpenAI's case—the Office of the National Cyber Director and the Office of Science and Technology Policy—reportedly worked closely with the company on the restricted release strategy.

Following Anthropic's Playbook

This approach mirrors Anthropic's voluntary constraint on its Claude Mythos model, which the company has kept under lock and key through a limited partnership program since April. Anthropic justified the decision by highlighting the model's potential for malicious use in cyber operations—particularly its ability to identify and exploit software vulnerabilities at machine speed, a capability that could devastate enterprise networks.

Generative AI systems have already proven useful for attackers: large language models can write malware and, in some cases, execute ransomware autonomously. Whether restricted-release models like Mythos pose a genuine threat or serve mainly as a marketing narrative remains contested—though the answer likely sits somewhere in the middle.

What Remains Unclear

Because these frontier cyber tools remain sequestered, their actual risk profile remains hard to assess. The restricted-release model aims to balance innovation with caution, but questions linger about whether government oversight and industry self-limitation represent sound policy or performative safety theater.