US Govt Push Delays OpenAI’s GPT-5.6 Rollout as Security Oversight Tightens

A planned wide public release of OpenAI’s newest frontier model has been slowed after U.S. government intervention, reflecting a sharper turn in Washington toward tighter oversight of advanced artificial intelligence systems. OpenAI said on Friday that it was postponing a full public launch of GPT-5.6 at the request of U.S. authorities, instead restricting early access to a small set of vetted partners whose identities have been shared with the government. The move signals rising national security concerns around frontier AI systems, as policymakers increasingly push for early visibility into powerful models before they are widely deployed. “Temporary step” as Washington builds new AI release framework In a blog post, OpenAI described the delay as a controlled, short-term measure tied to ongoing coordination with regulators. “We are taking this short-term step because we believe it is the strongest path to broader availability in the coming weeks, while we work with the Administration to develop the cyber Executive Order framework and a repeatable process for future model releases,” the company said. The company added that it had already briefed the government on the model’s capabilities before release and intends to continue working closely with official partners as it prepares for broader access. However, OpenAI also cautioned that this level of oversight should not become permanent, warning it could slow access for developers, businesses, cybersecurity professionals, and international collaborators who depend on advanced AI tools. Security concerns drive early access for US officials The decision comes amid growing concern in Washington over how frontier AI systems could be misused. U.S. policymakers are increasingly focused on identifying risks such as cyberattacks, automated hacking, and potential military applications before these tools reach the public. By reviewing models early, officials aim to evaluate threats in controlled environments and reduce the chance of harmful deployment at scale. The shift follows a broader policy direction under U.S. President Donald Trump, who recently signed an executive order establishing a voluntary framework that allows the government early access to “covered frontier models” before public release. GPT-5.6 “Sol” anchors new model lineup At the center of OpenAI’s upcoming releases is GPT-5.6 Sol , described as the company’s most advanced model to date. It is expected to be accompanied by mid-tier Terra and lower-cost Luna variants, designed to balance performance and accessibility across different use cases. The staged rollout reflects a broader strategy of tiered AI systems, where capability and cost vary across model versions to serve different segments of users and developers. Broader regulatory pressure across the AI industry The move comes as other major AI developers face increasing government scrutiny. Anthropic has also been ordered to restrict access to frontier models for certain foreign users, amid ongoing national security concerns. At the same time, both OpenAI and Anthropic have reportedly filed confidential paperwork for potential U.S. IPOs, while The New York Times reported that OpenAI may delay its public listing until next year. Together, these developments point to a rapidly evolving regulatory environment where advanced AI deployment is becoming closely intertwined with national security policy, corporate strategy, and global technology competition.