Trump Administration Asks OpenAI to Limit Release of GPT-5.6: Reports

President Donald Trump’s administration has asked OpenAI to slow down the initial release of GPT-5.6 to a small group of government-approved workers, while government officials test the model, according to reports of The Information and Axios.

The request is the second this month that the US government has intervened to limit the release of an AI version of the border, following the Anthropic Act. to stop the presence of people in Claude Fable 5 is Legend 5 for national security.

According to reports, the White House’s Office of the National Cyber ​​​​Director and the Office of Science and Technology Policy asked OpenAI to slow down the release of GPT-5.6 while the administration is developing a system to evaluate advanced AI models before they are widely deployed. Those familiar with the discussions say the request was driven by GPT-5.6’s “Mythos-like” abilities rather than a major change in AI principles.

The petition follows President Trump Executive Order earlier this month to instruct federal agencies to establish a voluntary testing system for advanced AI systems before they are rolled out, after weeks of internal debate over what the program should look like.

The move also marks a shift in the relationship between leading AI developers and Washington after years of developers demanding government regulation of the product.

During Senate testimony in 2023, OpenAI CEO Sam Altman urged lawmakers to implement it regulatory body for advanced AI systems, he says that self-monitoring may be necessary. Recently, Anthropic CEO Dario Amodei they argued that highly capable AI models should be subject to government-sponsored testing prior to deployment due to their potential to enable sophisticated cyberattacks, weapons tracking, and other national security threats.

The arguments have been fixed as Anthropic, OpenAIand Google have a published proposal that shows how the boundaries of AI should be managed. Although they differ in their approach, all three require systematic evaluation of the most capable models, transparency in safety testing, independent evaluation of high-risk systems, and a greater role for the government in managing the development of AI.

Managerial intervention can also test whether the strategies of leading AI companies can be applied in the same way across companies. Opponents warn that if major AI developers help create rules that are enforced inconsistently, then the limits of AI may be systematic seizure which favors a select group of companies while making it difficult for competitors to compete.