The US government has requested that OpenAI delay the release of its latest AI model, GPT 5.6, due to security concerns. The company will instead release a limited preview of the model to a small group of enterprise customers, with the federal government approving access on a case-by-case basis.
What Happened
According to reports from The Information and Reuters, OpenAI CEO Sam Altman informed employees that the company would put GPT 5.6 into a limited preview for select enterprise partners. During this preview period, the government will approve access customer by customer, as per The Information. This arrangement is more permissive than recent restrictions applied to Anthropic's Mythos 5 and Fable 5 models, which were subject to limitations on foreign access.
The request from the US government was reportedly made by the Office of the National Cyber Director and the Office of Science and Technology Policy. The delay in releasing GPT 5.6 is a significant development, as it highlights the growing concerns around AI security and the need for greater regulation in the industry.
Background and Context
The Trump administration's request to OpenAI follows a recent executive order signed by President Trump, which aims to create a voluntary framework for the federal government to vet powerful new AI models before they are released. This move represents a shift from the White House's previously deregulatory stance on AI, with Vice-President JD Vance stating last year that "excessive regulation of the AI sector could kill a transformative industry."
However, in light of rapidly improving model capabilities, including Anthropic's Mythos, which has been described as a "step up" over previous cutting-edge models by the UK's AI security body, the White House has changed its stance. The administration's concerns around AI security are likely driven by the potential risks associated with powerful models like GPT 5.6, which could be used for malicious purposes if not properly controlled.
Why It Matters to the Industry
The delay in releasing GPT 5.6 and the government's request for a staggered rollout represent a significant intersection of national security policy and AI deployment practice. For practitioners building on or deploying large models, this trend changes the timeline and scope for early integrations, third-party testing, and cross-border collaborations.
As major model releases increasingly trigger cross-functional reviews involving security, export-control, and national-security offices, companies rolling out frontier models face a mix of technical gating (limited previews, enterprise-only access) and policy gating (government-led customer approvals). This combined approach typically slows broad access, concentrates usage data in early partner cohorts, and raises operational overhead for access management and compliance.
What Comes Next
18,094 page views
