The release of OpenAI’s latest model GPT 5.6 will reportedly be different from previous releases. The company plans to share it only with a few close partners rather than distributing it to the public, as directed by the Trump administration, The Information reported.
At a meeting this week, CEO Sam Altman told staff that the government would “approve access on a customer-by-customer basis” during the preview period. Altman reportedly added that if the limited release goes well, OpenAI expects to do a general, widespread release in “a few weeks.”
In other words, the Trump administration appears to be pressuring OpenAI to do what Anthropic has already done voluntarily: keep its most powerful AI models secret.
Not only is OpenAI’s new model being considered by the government, but its staff also “worked closely” with the government on future releases, according to The Information. According to reports, the agencies that requested the limited release were the Office of the Director of the National Cyber Bureau and the Department of Science and Technology Policy.
The Trump administration initially said it would take a “hands-off” approach to AI, but in recent months it has pushed for federal oversight of new models. Earlier this month, President Trump signed an executive order directing certain AI companies to voluntarily submit new models to the government for testing and evaluation before making them public.
Earlier this year, Anthropic caused more than a little controversy when it announced that its new frontier cyber model, Claude Mythos, would be released to only a small number of partners through a program called Project Glasswing. Anthropic argued that its model was simply too powerful and could do more harm than good in the wrong hands. Observers have since debated whether Anthropic’s rhetoric was just a marketing gimmick or a legitimate attempt to prevent abuse of a powerful model. The answer may lie somewhere in between.
Cybercriminals have been using automated tools for a very long time, but in the age of generative AI they have more digital attack vectors than ever before. LLMs have proven adept at creating malware, with some being able to execute entire ransomware attacks autonomously.
A particular concern with cutting-edge cyber tools like Mythos is that these tools can identify and exploit software vulnerabilities at a speed that human analysts cannot match. This clearly poses a clear and significant problem for organizations running complex software infrastructures, as many software systems contain hidden bugs that act as entry points into corporate networks. However, these models are not publicly available, so it is difficult to know how much of a threat they actually pose.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
