Trump Administration Requests Delayed Release of OpenAI's GPT 5.6

The Trump administration has intervened in OpenAI’s product launch schedule, requesting that the company delay and stagger the release of its next AI model, GPT 5.6. This request, made on June 25, stems from concerns over cybersecurity risks that officials believe require further evaluation before the technology is made widely available. This action follows an executive order signed by President Trump on June 2, which established a voluntary framework permitting government cybersecurity teams up to 30 days to assess advanced AI models prior to their public rollout.

Discussions regarding the staggered release involved officials from the Office of the National Cyber Director and the Office of Science and Technology Policy. In response, OpenAI CEO Sam Altman informed staff that the launch would transition to a limited preview, accessible only to a select group of partners rather than an immediate full release.

OpenAI is not the only AI company facing scrutiny; Anthropic, known for its Claude model family, has also encountered restrictions on its advanced models due to similar national security evaluations. The executive order reflects a broader geopolitical context, particularly the intensifying US-China technology competition. Washington is concerned that advanced AI models could be exploited by adversarial nations or non-state actors if released without sufficient security vetting.

The voluntary review process established by the executive order represents a compromise, avoiding the more drastic measures of outright bans or mandatory pre-release government approval. An earlier draft of the order, which proposed stricter regulations, was postponed following significant pushback from the AI sector in May 2026.

This initiative appears to be aimed at fostering partnerships between the government and the AI industry, potentially leading to contracts as federal agencies look to integrate AI tools into defense, intelligence, and civilian operations. For the broader AI sector, the 30-day voluntary review window is seen as a less disruptive alternative compared to mandatory licensing or pre-market approval processes that have been suggested by some lawmakers.