OpenAI delays GPT-5.6 release at Trump administration's request
Government approval process keeps best AI tools from users and developers.
OpenAI has confirmed it is delaying the public release of its next-generation AI models, GPT-5.6, at the request of the Trump administration. The company will first share the models with a small set of customers preapproved by the US government, then work with the administration to slowly expand access. OpenAI expressed dissatisfaction with the delay, calling it temporary, and stated it hopes to make GPT-5.6 available to everyone in the coming weeks. The move follows a recent executive order from President Trump addressing cybersecurity concerns of powerful new AI models, which outlined a voluntary process for labs to share models with the government 30 days before broader release. However, no such framework exists yet, leaving frontier AI labs in an uncertain period where government collaboration seems less than voluntary.
The White House's request comes after similar actions against Anthropic, which was asked to limit access to its most advanced models. OpenAI's GPT-5.6 will be available in three flavors: Sol (most capable, excelling in cybersecurity, biology, and agentic benchmarks), Terra (mid-tier), and Luna (fast and affordable). The company has implemented a 'layered safeguard stack' to prevent malicious use, such as cyberattacks. The administration continues to collaborate with frontier AI labs, but the situation creates an uncertain environment for US AI innovation and competitiveness with China.
- OpenAI delays GPT-5.6 public launch at US government request, citing cybersecurity concerns.
- Models come in three tiers: Sol (top-tier), Terra (mid), Luna (fast/cheap) with a new safeguard stack.
- Voluntary government review process not yet defined, leaving labs in a compliance gray area.
Why It Matters
Government interference could slow AI innovation and set a precedent for future model releases.