The White House has requested OpenAI to restrict the initial release of its upcoming GPT-5.6 model due to safety concerns, according to medianama.com. OpenAI plans to provide access only to a small group of partners during an initial preview, with CEO Sam Altman stating that the government will approve access on a "customer by customer" basis. A wider release is expected if the rollout proceeds without issues.
This move follows a similar directive issued on June 12, when the US government ordered Anthropic to suspend access to its flagship AI models, Fable 5 and Mythos 5, for all foreign nationals worldwide. Anthropic complied but challenged the basis of the order, arguing that the cited jailbreak vulnerability was limited and previously known. The government’s concerns over potential safety risks have led to increased oversight of advanced AI systems, with OpenAI’s rollout now subject to government approval.
The government’s intervention marks a continuation of heightened regulatory scrutiny on AI model releases, especially those with significant capabilities. The Anthropic order has set a precedent for controlling access to frontier AI technologies amid national security and safety concerns. OpenAI’s GPT-5.6 rollout under government supervision reflects broader efforts to manage risks associated with advanced AI, similar to controls seen in other high-impact technology sectors.
The legal challenge to the Anthropic export control directive was initiated on June 23 by legal technology company Legion Le, highlighting ongoing disputes over the scope of government authority in AI regulation. OpenAI’s phased GPT-5.6 release remains contingent on government approval, with the initial limited preview expected to begin shortly, according to medianama.com.