WASHINGTON, July 8 (DC Times Online) — OpenAI is not releasing its newest AI model to everyone right away. Instead, the company says GPT-5.6 Sol is going first to a small group of trusted partners while the Trump administration reviews possible cybersecurity risks.
The move matters because it shows how Washington is trying to slow down the release of powerful AI tools until officials are more comfortable with the risks. For most users, the immediate effect is simple: the model is not broadly available yet.
What is OpenAI releasing?
OpenAI says Sol is its newest model. In statements cited by several outlets, the company described it as its strongest model yet and said it is better at helping people find and fix vulnerabilities than at carrying out cyberattacks.
OpenAI also said the model does not cross its internal “Cyber Critical” threshold. That is the company’s own line for when it thinks a model raises especially serious cyber risk.
Why is the rollout limited?
The White House asked OpenAI to slow the launch, according to OpenAI’s statement and reporting on the administration’s talks with the company. The concern is that advanced AI could be used to speed up hacking, probe weak points in software, or threaten systems that matter to the economy and national security.
That is why the story is centered in Washington. The decision ties into a broader federal effort to look at the risks of frontier AI, meaning the most advanced models.
What rule is Washington using?
The Trump administration’s June executive order created a voluntary federal review framework for advanced AI models before release. The White House framework and fact sheet describe a process that can take up to 30 days and is meant to help agencies assess threats to financial systems, national security, and other critical systems.
The Office of the National Cyber Director and the Office of Science and Technology Policy are among the offices named in the reporting as part of that review process.
What does a “limited preview” mean?
OpenAI says access is starting with a small group of trusted partners. The company said those partners’ participation has been shared with the government.
OpenAI has not named the customers approved so far. ABC News reported that the number is about 20.
The company says the preview is temporary and that broader access is expected in the coming weeks. OpenAI also said it does not believe this kind of government access process should become the long-term default.
What does this mean for users and businesses?
For most people, it means waiting. Fox News reported that the model has not yet been made available in ChatGPT.
For developers, businesses, and cybersecurity teams, the limited rollout means the model’s most advanced features will stay behind a preview process for now. That can slow testing, product planning, and any work that depends on the newest OpenAI tools.
For the federal government, the bigger question is whether this becomes a standard step for future AI launches. If Washington keeps asking for pre-release reviews of frontier models, companies may have to build that delay into their release plans from the start.
