OpenAI Delays Launch of Powerful GPT-5.6 Models at White House's Request

On June 26, OpenAI announced that, at the request of the Trump administration, it would delay the public release of its latest frontier AI models, the GPT-5.6 series (which includes Sol, Terra, and Luna). Despite these models' potential value in cybersecurity defense, OpenAI stated that it would first provide a limited preview to "a small group of trusted partners," with the partner list shared in advance with the U.S. government. The company's statement noted that the original plan had been to brief the government on the models' capabilities ahead of release, with a gradual public rollout planned for "the coming weeks."

This decision comes just two weeks after Anthropic was forced to take its most advanced AI model offline due to export control directives from the White House. According to WIRED, the disagreement between Anthropic and the White House remains unresolved, and some employees are still barred from using the company's top-tier model. OpenAI's delay highlights the U.S. government's ongoing pressure on AI safety and export controls, particularly through stricter approval processes for cutting-edge models with dual-use capabilities in cyber offense and defense.

Analysts note that this postponement is not a permanent ban but a phased-access strategy. OpenAI emphasized its commitment to "broad access," but current priorities have shifted toward aligning with national security reviews. As AI capabilities grow exponentially, balancing innovation speed with safety oversight will remain a core challenge for both the industry and policymakers. Whether the GPT-5.6 series will be fully launched as planned in the coming weeks remains to be seen.