OpenAI Delays GPT-5.6 Series Launch at White House Request

OpenAI announced on June 26 that it is delaying the public deployment of its latest frontier AI models, the GPT-5.6 series (codenamed Sol, Terra, Luna), at the request of the Trump administration. Despite these models' potential to assist cybersecurity defenders, the company decided to first offer a "limited preview" to a small group of trusted partners known to the government, with plans to gradually expand access. In a statement, OpenAI said: "We believe in broad access and plan to fully open the GPT-5.6 series within the coming weeks."

CEO Sam Altman confirmed in an internal memo that OpenAI had demonstrated the capabilities of GPT-5.6 Sol, Terra, and Luna to federal officials, and the White House formally requested a phased rollout due to cybersecurity concerns. Under the arrangement, initial preview access is limited to enterprise customers that have been evaluated and approved by the government, with each collaboration requiring reporting. This move follows competitor Anthropic's forced disabling of its advanced models due to strict export bans, prompting industry experts to call for a more consistent and transparent federal framework that balances national security with AI development.

Although OpenAI emphasized its "belief in broad access," this delay highlights the U.S. government's increasing regulatory oversight of cutting-edge AI capabilities. Industry analysts suggest that the White House's request for a phased release could become the new normal for future frontier models. Notably, the dual-use nature of these models in cybersecurity—capable of both aiding defenders and being exploited by attackers—will push policymakers to find a finer balance between technological innovation and security risks.