OpenAI’s most powerful AI model limited to 20 US-approved partners
▼ Summary
– OpenAI has released GPT-5.6 Sol, its most powerful model, to approximately 20 partners individually approved by the US government.
– This is the first time an American AI company has launched a frontier model under a government-managed access list.
– The release goes beyond the voluntary pre-release review framework established by Trump’s AI executive order.
OpenAI has quietly rolled out its most advanced artificial intelligence system, GPT-5.6 Sol, to a tightly controlled group of roughly 20 organizations, each pre-approved by the U.S. government. This marks the first time an American AI firm has released a frontier model under a formal, government-managed access list, moving beyond the voluntary pre-release review framework introduced by Trump’s AI executive order.
The move signals a major shift in how the U. S. is managing the most powerful AI technologies. Instead of relying on companies to self-regulate, the government now directly determines who can access cutting-edge AI systems. The approved partners include a mix of defense contractors, critical infrastructure operators, and select research institutions, though OpenAI has not disclosed the full list.
GPT-5.6 Sol represents a significant leap in capability, with performance benchmarks that far exceed previous models. OpenAI has described it as a “safety-first” release, but critics argue that limiting access to such a small number of vetted entities could stifle innovation and concentrate power in a handful of government-linked organizations. The company maintains that the restrictions are necessary to prevent misuse, particularly in areas like cybersecurity, bioweapons development, and autonomous systems.
The approval process, overseen by the Department of Commerce and the National Security Council, required each partner to demonstrate robust security protocols and a clear, beneficial use case. OpenAI also agreed to ongoing monitoring and reporting requirements, a condition that goes beyond typical commercial agreements.
This approach stands in contrast to the more open release strategies of previous models. OpenAI has faced mounting pressure from lawmakers and national security officials to prevent its technology from falling into the hands of adversaries or being used for harmful purposes. The limited preview model is seen as a compromise between rapid deployment and responsible oversight.
Industry observers note that this could set a precedent for other AI companies. If the government continues to assert control over frontier model access, the landscape for AI development and deployment could shift dramatically, with national security interests increasingly shaping the commercial availability of the most advanced systems.
(Source: The Next Web)
