The Trump administration has asked OpenAI to stagger the release of its new "GPT-5.6" model, first offering a preview version to a small set of trusted, government-approved customers before opening it to the public weeks later if no problems emerge.
June 25, 2026 · AI Policy
The U.S. Government Wants to Vet Who Gets GPT-5.6
The Trump administration has asked OpenAI to stagger the release of its next model, citing security concerns — starting with a small set of trusted partners and expanding only after the government approves access customer by customer.
GPT-5.6
Next OpenAI model at the center of the request
2
Federal offices behind the ask: National Cyber Director & OSTP
~2 wks
Earlier, Anthropic pulled a frontier model under regulatory pressure
The staggered rollout, step by step
STAGE 1
Limited preview
A short list of trusted partners only.
→
STAGE 2
Per-customer vetting
Government approves access customer by customer.
→
STAGE 3
Wider rollout
Broader release in the coming weeks.
Welcomed
Some developers welcome the stronger security emphasis and the state directly managing national-security risks of frontier models.
Questioned
Others see it as restricting public access to advanced models — a shift from voluntary safety frameworks that could curb commercial freedom.
Also moving this week
Google vs. publishers
A tough negotiating stance — pressing publishers in an AI pilot to provide content for training, with consequences hinted for those who decline.
Agent orchestration
A "digital workforce" of ~20 agents in a hub-and-spoke setup, with a Chief of Staff agent steering specialized ones via custom commands.
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…