OpenAI is limiting access to its latest AI model, GPT-5.6 Sol, to a small group of government-approved partners as part of a temporary cybersecurity review ordered by the Trump administration. The company said the restricted release is a temporary measure on the path to broader availability in the coming weeks, following an executive order signed in June that established a framework for federal vetting of advanced AI systems before public release.

Why Is the Government Reviewing OpenAI's New Model?

The Trump administration's decision to vet GPT-5.6 Sol reflects growing concerns about how powerful AI systems could be misused by malicious actors. Earlier in June, officials became alarmed after Anthropic, OpenAI's rival, warned that its Mythos model could be weaponized to find software vulnerabilities in ways that threaten critical computer networks. This warning prompted the government to take action against Anthropic, which took two models offline to comply with a Trump directive.

The executive order established a voluntary framework allowing the federal government to vet the national security risks of the most advanced AI systems for up to 30 days before their public release. OpenAI acknowledged that while its Sol model "is better at helping people find and fix vulnerabilities" than at carrying out cyberattacks, there could be unforeseen risks, especially if the model is combined with other tools.

"That uncertainty, along with the model's broader step change in capabilities, is why we are pairing the model's increased capabilities with stronger safeguards and a phased release," OpenAI said in a statement.

OpenAI, Statement on GPT-5.6 Sol Release

How Does This Compare to Anthropic's Experience?

OpenAI's cautious approach contrasts sharply with the more contentious relationship between the Trump administration and Anthropic. The Pentagon designated Anthropic as a national security risk for raising ethical and safety concerns about AI usage in warfare, and Trump ordered federal agencies to stop using Claude, Anthropic's chatbot. Anthropic responded with a lawsuit that is still working through federal courts.

OpenAI has taken a more collaborative approach. CEO Sam Altman spoke with U.S. Commerce Secretary Howard Lutnick about the model release, part of a series of negotiations between AI industry executives and Trump officials. The company has not named any of the roughly 20 customers approved to use the new model so far.

What Are Cybersecurity Experts Saying About Government Intervention?

A broad group of cybersecurity experts has criticized the government's actions, particularly the decision to force Anthropic to shut down its Fable model, which the company had pitched as a safer version of Mythos. The model has been unavailable for two weeks.

"I just want to say that pretty much nobody in the cybersecurity industry believes that there's any factual basis for this action," said Alex Stamos, a Stanford University cybersecurity expert and chief product officer at AI security company Corridor.

Alex Stamos, Chief Product Officer at Corridor and Former Chief Security Officer at Meta

Stamos reviewed an analysis of Anthropic's research on Fable by Amazon, Anthropic's primary cloud computing backer, and found no risks that aren't present with other publicly available AI models, including those made in China. He warned that government intervention could undermine U.S. competitiveness in the global AI race.

Steps to Understanding the Government's AI Vetting Framework

• Voluntary Participation: The Trump administration's framework invites AI developers to participate voluntarily in a 30-day national security review before releasing advanced systems to the public.
• Phased Release Strategy: Companies like OpenAI are implementing staggered releases to a small group of trusted, government-approved partners before broader availability, allowing time for security testing.
• Risk Assessment Process: Developers must evaluate whether their models could be weaponized or combined with other tools to pose cybersecurity threats to critical infrastructure.
• Transparency Requirements: Companies are expected to communicate with government officials about their models' capabilities and potential risks, as OpenAI did with Commerce Secretary Lutnick.

What Does This Mean for OpenAI's Future?

OpenAI's statement that it does not believe "this kind of government access process should become the long-term default" suggests the company views the current vetting as temporary. The company expects to expand access to GPT-5.6 Sol in the coming weeks once the cybersecurity review is complete.

However, the precedent set by this review could reshape how advanced AI systems are released in the United States. If the government's framework becomes standard practice, it may slow the pace at which companies like OpenAI can bring new models to market, potentially affecting their competitive position against international rivals.

The situation also highlights tensions between national security concerns and the desire to maintain U.S. leadership in AI development. As the government tightens oversight, companies must balance compliance with the need to innovate quickly in a rapidly evolving field.