White House to individually approve customer access to GPT-5.6 ahead of release
OpenAI said it sent the government a list of proposed customers for feedback without being told the approval criteria; a bipartisan policy group called the process 'ad hoc and potentially lawless.'
- Government & policy
- Models & capabilities
- Notable
Reporting on OpenAI’s restricted preview of GPT-5.6 Sol detailed the specific mechanism behind the restriction: rather than issuing a blanket clearance for the model, the White House was approving access on what The Decoder described as a “customer by customer basis.” OpenAI said it submitted a list of proposed partner organisations to the government and received feedback, but could not disclose the criteria being used to approve or reject individual customers. The push for a phased release reportedly came out of discussions with the Office of the National Cyber Director and the Office of Science and Technology Policy, with the Commerce Department separately signalling it wanted broader agency sign-off before any wider rollout.
The arrangement drew criticism for how far it stretched a nominally voluntary framework. Executive Order 14409, signed on 2 June, had established early government access to frontier models before public release as voluntary; in practice, OpenAI had limited room to decline a request tied to a presidential order. Brad Carson, head of the bipartisan group Public First, told CNN the approach was “ad hoc and potentially lawless,” pointing to the absence of any formal legal framework governing which customers could be approved or on what basis. Sam Altman said OpenAI had made clear to the government that the arrangement was “not our preferred long term model” and that it was working toward a more sustainable process for future releases.