OpenAI to limit GPT 5.6 release under White House pressure

OpenAI plans to share its new artificial intelligence model, GPT 5.6, only with a select group of close partners instead of the public, according to a report by The Information. At a company meeting this week, CEO Sam Altman reportedly told staff that the US government would be “approving access customer by customer” during an initial preview period. Altman reportedly added that if this limited release goes well, OpenAI hopes to follow with a general, broader release a couple of weeks later. The shift in strategy comes as OpenAI staffers reportedly worked closely with the government on the upcoming release, marking a departure from the company’s typical public launch format.

The Trump administration appears to be pressuring OpenAI to keep its most powerful AI models under wraps. According to The Information, the specific US government agencies that requested the limited release are the Office of the National Cyber Director and the Office of Science and Technology Policy. This intervention represents a notable shift for the administration, which had previously positioned itself as taking a hands-off approach to AI regulation. However, earlier this month, Trump signed an executive order directing certain AI companies to voluntarily submit new models to the government for testing and evaluation before releasing them publicly.

The administration’s cautious stance mirrors voluntary restrictions already established by other industry players, though OpenAI’s restrictions are driven by federal pressure rather than a voluntary decision. Earlier this year, Anthropic announced that its Claude Mythos model would only be released to a small group of partners through a program called Project Glasswing. Anthropic argued that the model was “too powerful and could, in the wrong hands, cause more harm than good.” The debate highlights growing concerns over generative AI, with some warning that the technology provides cybercriminals with “more digital ammunition than ever before” to write malware and execute autonomous ransomware attacks. These concerns are particularly acute for frontier cyber tools, which are capable of identifying and exploiting software vulnerabilities at speeds that human analysts cannot match.

Why it matters

The Trump administration’s pressure on OpenAI signals a significant shift toward direct federal oversight of AI model releases, moving away from a previously hands-off approach. This transition from voluntary industry standards to active government intervention could redefine how major AI developers deploy their technology.