The Trump administration has reportedly asked OpenAI to delay the full public launch of its newest frontier AI model, GPT-5.6, because of national security concerns. Instead of making the model available to all users at once, OpenAI has agreed to release it in phases, starting with a small group of trusted organizations, reports The Information.
During this early rollout, every company or institution that wants access will need approval from the US government. This is the first known case where the US government has directly influenced the release schedule of a major commercial AI model before its public launch.
The request came from the White House’s Office of the National Cyber Director (ONCD) and the Office of Science and Technology Policy (OSTP), which have been working on a new system to review advanced AI models before they are widely released. US officials believe the latest generation of AI has become powerful enough to affect national security, cybersecurity and critical infrastructure.
Their main concern is that highly capable AI models could be misused to discover software vulnerabilities, create advanced malware, automate cyberattacks or help hostile governments and criminal groups carry out sophisticated cyber operations. Rather than waiting until such models are already available to millions of users, the administration wants to examine them first and decide whether extra safety measures are needed.
According to the report, OpenAI CEO Sam Altman told employees that the company had agreed to cooperate with the government’s request, even though it would have preferred a normal public launch. GPT-5.6 will first be released as a limited preview, with reports suggesting that around 20 enterprise customers could receive early access while government agencies review the model’s real-world performance.
If no major issues are found, OpenAI expects to expand access within a few weeks. The company has made it clear that this phased rollout is only temporary and is not expected to become its standard approach for future AI releases.
The government’s request also follows a broader change in US AI policy. Earlier this month, President Donald Trump signed an executive order that encourages companies developing frontier AI systems to give the federal government early access to their models before public release. The goal is to allow security experts to test these systems for potential risks before they reach millions of users.
OpenAI claims that GPT-5.6 has already passed extensive internal safety testing and does not exceed the company’s highest-risk thresholds for dangerous cyber capabilities. According to the report, the company used more than 700,000 GPU hours to test the model, along with external expert evaluations and cybersecurity assessments. GPT-5.6 is part of a broader family of AI models that includes Sol, the flagship frontier model, along with Terra and Luna, which are designed for different performance and efficiency needs.
The decision also comes after increased government scrutiny of other leading AI companies. Earlier this month, Anthropic paused access to its advanced Fable 5 and Mythos 5 models following concerns from U.S. officials about their powerful cybersecurity capabilities and the possibility of foreign misuse.
The Tech Portal is published by Blue Box Media Private Limited. Our investors have no influence over our reporting. Read our full Ownership and Funding Disclosure →
Ashutosh is a Senior Writer at The Tech Portal, largely reporting on new tech, and intersection of technology and business. Ashutosh’s career spans across nearly a decade of technology writing across multiple platforms and languages.
