OpenAI must initially release its new GPT-5.6 model only to partners approved individually by the U.S. government, rather than launching broadly. This stems from the Trump administration's push for cybersecurity review of advanced AI models and mirrors a recent incident where the government pulled Anthropic's Fable model offline after public release. CEO Sam Altman said OpenAI does not want this per-customer approval to be a permanent arrangement and will work toward a sustainable long-term approach.
Summaries like this, in your inbox every morning.
Sign up free →What happened
OpenAI must initially limit access to its new GPT-5.6 model to a small group of partners at the U.S. government's request, CEO Sam Altman announced during an internal Q&A. The government will approve access on a customer-by-customer basis during the preview phase, with a broader release expected a couple of weeks later if all goes smoothly. The push came from talks with the Office of the National Cyber Director and the Office of Science and Technology Policy.
Why it matters
This move reflects the Trump administration's executive order calling for voluntary review of new AI models around cybersecurity. Altman stated OpenAI does not prefer this long-term model and will work with the government toward a more sustainable approach—suggesting the company views the per-customer approval as a temporary constraint. The case may indicate that the government's framing of AI review as 'voluntary' is increasingly accompanied by direct intervention.
What to watch
The timeline hinges on government sign-off; Altman hopes for broader release in a couple of weeks. Commerce Secretary Howard Lutnick called Altman to warn against moving forward without sign-off from more agencies, signaling government coordination at the Cabinet level. This follows a similar incident with Anthropic, whose Fable model was forced offline by the U.S. government after launch, despite prior collaboration on security testing.
No comments yet. Be the first to share your thoughts!
Log in to join the discussion
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
5 minutes a day. The AI essentials.
200+ sources · Email / LINE / Slack
