Breaking

US Government Asks OpenAI to Stagger Release of New GPT-5.6 Model

June 25, 2026 at 17:52 EDT

Foundation Models
Policy & Regulation

The Trump administration has asked OpenAI to roll out its new frontier model, "GPT-5.6," in stages over security concerns. In response, OpenAI told staff it would first offer the model as a limited preview to select partners, with the government approving access on a customer-by-customer basis.

June 25, 2026 · OpenAI · U.S. Government

Washington Asks OpenAI to Stagger Its Next Flagship — and OpenAI Will Comply

Over national-security and cyber concerns, GPT-5.6 will launch as a limited preview — with the government approving access one customer at a time before any broad release.

~30 days

Window agencies may take to run cybersecurity evaluations before a frontier model ships

1×1

Access granted customer by customer, with government sign-off on each

5.6

The flagship version at issue — successor to GPT-5.5, "a meaningful improvement"

The release path: a new normal for frontier launches

Step 1

Government preview

Cyber evaluation by national security & science offices

→

Step 2

Limited rollout

Trusted partners only, approved one customer at a time

→

Step 3

Broader release

Timing and scope still undetermined

Seen as reasonable

Staggering allows real security vetting of a more capable model
A voluntary framework OpenAI says it will cooperate with

Concerns raised

Approval process adds delay; Altman calls it "not the preferred long-term model"
Risk of slowing development and eroding edge over competitors

What's still unknown: parameter count, context length, benchmark scores, and full-release timing are all undisclosed. No official OpenAI statement is confirmed — an internal memo from Sam Altman is the primary account.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.