OpenAI Has New Types of AI. Here’s Why You Can’t Use It

OpenAI is slow public release of the next generation of AI models, GPT-5.6, at the request of Trump’s White House, the company confirmed on Friday. OpenAI it said it would share first examples and a few customers, who will already be approved by the US government. Will work with management to gradually increase access.

OpenAI is not happy about this, according to a person familiar with the company’s views, but believes that the delay in government approval is temporary. In a blog post, the company said it hopes to make GPT-5.6 available to everyone in the coming weeks. OpenAI’s plans to delay the next generation of AI models at the request of the Trump administration were the first report by The Information.

“We do not believe that such government access measures should be temporary,” OpenAI wrote on its blog. “It preserves the best tools from users, developers, businesses, cyber defenders, and partners around the world who need them. We are taking this short step because we believe it is the most powerful way to achieve greater availability in the coming weeks, while we work with the Ministry to develop a cyber Executive Order and an iterative process to release future models.”

Earlier this month, President Trump signed it Executive Order which aims to address the cybersecurity concerns of powerful AI models. The order said the White House would create a “voluntary process” for AI labs to share their samples with the government 30 days before release. The project also included drawings, saying that the US government will not turn its voluntary approach into a de-facto licensing regime for the production of AI models. But in a statement on Friday, OpenAI executives said there are no voluntary measures in place. As a result, border AI labs are in a very strange situation, where working with the US government on the implementation of your AI model does not seem to be voluntary.

The White House is asking OpenAI to destroy the release of its AI models just two weeks after it was released Export control directive to Anthropicwhich led the company to take its advanced AI models online for all customers. The Anthropic Debate with the White House is it is still not resolvedand some of the company’s employees are still restricted from using its most advanced AI models.

The Trump Administration’s request for OpenAI and Anthropic to limit access to their most advanced AI models Uncertain location of other US AI labs. Over the past two years, the Trump administration has sought to streamline regulations and red tape that could hinder America’s AI capabilities, as well as damage the country’s competitiveness with China. However, in recent months, the White House has become increasingly concerned about the cybersecurity capabilities of new AI models, and has made efforts to address the issue.

OpenAI plans to expand the number of customers who can share GPT-5.6 next week, including international partners. OpenAI executives said they could not share details about how the White House approves these clients — the company only sends the US government a list, and then gets responses, the officials said.

The White House did not immediately respond to WIRED’s request for comment.

OpenAI says its GPT-5.6 AI models will come in three versions: Sol, its most capable model version; Terra, the central color of the model; and Luna, the fastest and cheapest version. The company says the GPT-5.6 Sol is its most reliable model for cybersecurity, biology, and human resources tests. Along with the new technology, OpenAI says it has “layered security,” which is intended to prevent malicious actors from using its AI model for cyber attacks, among other malicious practices.