White House Will Ad Hoc Decide Who Can Individually Access GPT-5.6
Zvi's AI Roundups · Zvi Mowshowitz · 2026-06-26
Zvi Mowshowitz argues that the Trump administration's customer-by-customer approval process for GPT-5.6 access is a maximally bad AI policy that will slow deployment, widen the public-private AI capability gap, and create competitive vulnerabilities against China.
Appears in
Extraction
Topics: ai-policyfrontier-ai-regulationgpt-5-6us-government-aiopenai
Claims
- The Trump White House is approving GPT-5.6 access on an individual customer-by-customer basis in an opaque, ad hoc manner.
- This policy slows AI deployment but not development, steadily widening the gap between what labs have internally and what is publicly available.
- A staggered Western release schedule could allow Chinese models, currently approximately nine months behind, to close the capability gap.
- The current policy is a direct consequence of the anti-regulation camp blocking earlier, more rational regulatory frameworks when the opportunity existed.
- Anthropic will face disproportionately more scrutiny than OpenAI under the new ad hoc approval system.
Key quotes
Ad hoc opaque politicized decisions from the White House on who gets frontier intelligence, however, is Not The Way. It is maximally Not The Way.
We went from zero AI regulation to CFIUS-but-for-API-access in about a week.
The gap between what is available to the public, and what the labs have internally, will steadily widen from this day forward.