OpenAI's newest flagship AI model, GPT-5.6 Sol, is launching with an unprecedented gatekeeper: the U.S. government. The company announced that the model will initially be available only to a small group of trusted partners approved by the Trump administration, marking a significant shift in how cutting-edge artificial intelligence tools are released to the public.
OpenAI said it previewed Sol's capabilities with the U.S. government ahead of launch and, at the government's request, is starting with a limited rollout to partners whose participation has been shared with federal officials. The model, part of OpenAI's new GPT-5.6 lineup that also includes Terra and Luna, is described as the company's strongest yet, with improvements in coding, biology, and cybersecurity tasks.
The cybersecurity capabilities are what have drawn Washington's close attention. OpenAI says Sol is better at helping users find and fix software vulnerabilities than reliably carrying out full cyberattacks. The company also states that Sol does not cross its internal «Cyber Critical» threshold, though it acknowledges that benchmark tests cannot predict every possible use when the model is combined with other tools.
The Trump administration has been building a new AI cybersecurity framework. A June executive order states that advanced AI can strengthen the country but also creates national security concerns requiring coordination across federal agencies and private companies. The White House fact sheet says the order creates a voluntary framework for covered frontier models and allows the government to secure early access for trusted partners, while clarifying that it does not authorize mandatory licensing, pre-clearance, or permits for releasing AI models.
This distinction is important. The administration says the approach is about security. Critics worry it could evolve into quiet government control over who gets the newest AI tools first. The limited preview means that, for now, the bigger question may not be when consumers can use the model, but who gets access first and how those decisions are made.
OpenAI is not the only AI company to face government pressure over powerful models. On June 12, Anthropic said the U.S. government directed it to suspend access to its Fable 5 and Mythos 5 models by foreign nationals. Anthropic said the order forced it to disable the models more broadly to comply. The U.S. Commerce Department later lifted export controls on those models, and Anthropic said it would restore access after agreeing to strengthen safeguards, work with the government on model-release protocols, and report malicious activity tied to the models.
The Anthropic episode provides a useful comparison for OpenAI's limited rollout. In both cases, the issue is no longer only about what an AI model can do. It is also about how quickly powerful AI tools should be released, who should get early access, and what role the government should play.
As of early July, OpenAI says GPT-5.6 Sol is not available in ChatGPT during the preview, and the company has not announced a general-availability date. The model will reportedly be accessible only to customers approved by the Trump administration during this initial preview period. OpenAI said GPT-5.6 models will initially be available through the API and Codex to a select group of trusted partners and organizations, with broader availability planned for ChatGPT, Codex, and the API soon.
The cybersecurity implications are central to the debate. Security teams want better tools because they are already stretched thin. Hospitals, healthcare systems, banks, local utilities, and businesses all run on software that can have hidden vulnerabilities. A stronger AI model could help find those weak spots before criminals exploit them. However, the same AI that can help a security team fix a software flaw could also give scammers and hackers a head start if the guardrails fail.
For now, the debate comes down to balancing security, access, and accountability as these AI tools move closer to the services people rely on every day. AI tools are getting more capable, and that means scammers will keep using the buzz around them to trick users. Experts advise caution with emails, texts, or ads that promise early access to GPT-5.6 Sol or any restricted AI model, recommending that users go directly to official company sites instead of clicking links.
