The recent limited release of OpenAI's advanced GPT-5.6 models, including Sol, Terra, and Luna, to a select group of "vetted partners" has drawn sharp criticism from AI policy experts, who describe the process as a "de facto licensing regime" lacking transparency. Miles Brundage, a prominent voice in AI governance and former OpenAI policy lead, voiced strong concerns on social media, stating, > "Re: why it's (morbidly) hilarious, we are in day N of a de facto licensing regime with literally no public information about the actual criteria or process involved, and constant gaslighting about how it's not mandatory when it obviously is."
This controlled rollout follows a request from the U.S. government, specifically the Trump administration, to assess potential national security risks associated with frontier AI models. An executive order signed by President Trump earlier this month established a voluntary framework, allowing the government up to 30 days of early access to advanced AI systems for review. Critics, however, argue that this "voluntary" measure is effectively becoming mandatory for major developers.
The process has been widely criticized for its opacity and potential to stifle innovation. Dean Ball, a former Trump AI adviser, labeled the opaque review an "increasingly draconian" measure, warning of unchecked government power over market access without clear standards or recourse. Stanford's Alex Stamos questioned the factual basis for restricting GPT-5.6, suggesting there is no unique risk posed by these tools compared to existing AI, including systems developed by Chinese entities.
OpenAI has confirmed its cooperation with the government's request, stating the preview is intended to last only weeks before broader availability, pending a new cybersecurity-focused executive order framework. The company maintains that the enhanced capabilities of models like Sol, which offer advanced coding and complex problem-solving, can ultimately benefit defensive applications more than offensive ones. Despite short-term cooperation, OpenAI has emphasized that the current government access process should not become the long-term standard for AI development and deployment.
The situation highlights a growing tension between national security concerns and the principles of open innovation in the rapidly evolving field of artificial intelligence. Experts fear that without clear, publicly defined criteria and processes, such unofficial regulatory mechanisms could disadvantage U.S. competitiveness and create an uneven playing field for AI developers.
