The Trump administration is expanding its involvement in the release of frontier AI models built by OpenAI and Anthropic, asking companies to stagger public launches or temporarily cut off access, citing national security and export controls.
June 2026 · Frontier AI Oversight
Washington Steps Into the Release of Frontier AI Models
The Trump administration is asking OpenAI and Anthropic to stagger launches or cut off access to their most advanced models — citing national security and export controls after a June 2 executive order.
June 2
Executive order signed — no mandatory tests, but "voluntary" model access required
30-day
Government review window granted for frontier models before release
2
Frontier labs already directly affected — OpenAI & Anthropic
Two labs, two government actions
OpenAI
Model: GPT-5.6
Asked to stagger the launch — begin with government-approved partners only, delaying general availability.
Anthropic
Models: Fable 5 / Mythos 5
All access briefly suspended under a Commerce export directive — restored after negotiations.
A shift in stance — the road to oversight
Early 2026
"Hands-off" stance toward AI regulation
→
~May 2026
Course change over cyber, export & foreign-access concerns
→
June 2026
EO + direct requests to stagger and restrict releases
The security case
Cybersecurity risks, export controls and foreign-national access justify vetting frontier models before wide release.
The industry worry
Voluntary review could act like a de facto licensing regime — delaying launches and abruptly cutting off users.
Benchmarks and pricing for GPT-5.6 and the affected models remain undisclosed — the concrete impact will only become clear with future releases.
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…