OpenAI Plans Limited GPT-5.6 Rollout After White House Request
The planned restricted release marks an early test of Washington’s role in reviewing powerful AI models before they reach wider commercial use.
News
- India-US Talks Put AI Stack In Focus
- OpenAI Plans Limited GPT-5.6 Rollout After White House Request
- Employers Still Want MBAs Even as One in Three Hands Entry-Level Work to AI
- AI Dispatch | Infosys Pitches AI Growth
- India May Take Minority Stake In Sarvam AI
- India Eases Satcom Rules by Dropping Sourcing Clause
OpenAI plans to limit the initial rollout of its upcoming GPT-5.6 model to a small group of government-approved partners following a request from the White House in an early test of Washington’s emerging approach to frontier oversight.
OpenAI Chief Executive Sam Altman told employees the company would release GPT-5.6 in a limited preview, The Information first reported, with the government approving access customer by customer during the initial period.
Axios reported that the White House’s Office of the National Cyber Director and Office of Science and Technology Policy asked OpenAI to limit the rollout while the administration builds a framework for evaluating the security of new models.
The model has not been publicly announced by OpenAI.
Altman told employees that the company hopes to move to a broader release within a couple of weeks if the preview goes as planned, according to The Information. He also said OpenAI had made clear to the US government that this was not its preferred long-term model for releases.
The request follows US President Donald Trump’s 2 June executive order on advanced AI innovation and security, which calls for closer cooperation between the federal government and AI developers as model capabilities raise cyber and national-security concerns.
The OpenAI case also follows a sharper government intervention involving Anthropic. On 12 June, Anthropic said the US government had issued an export-control directive requiring it to suspend access to its Fable 5 and Mythos 5 models for foreign nationals. Anthropic said the order effectively forced it to disable the models for all customers while it complied.
The two cases point to a more active, if still uneven, US approach to frontier AI releases. The Trump administration has said it wants to keep American AI moving quickly while protecting critical systems from cyber threats. In practice, the government is now beginning to scrutinize specific models before they reach wider commercial use.
That leaves AI companies with a difficult balance: moving fast enough to compete with rivals, including open-source models from China, while satisfying officials who worry that the most capable systems could be used to identify vulnerabilities, automate cyber operations or aid hostile actors.
For OpenAI, the immediate question is whether the GPT-5.6 preview remains a short delay before wider access or becomes a template for future frontier model launches.
