OpenAI Clears Safety Review for GPT-5.6 Broad Rollout

OpenAI completed a voluntary safety review with the U.S. Department of Commerce in early July 2026, clearing the path for a broad public rollout of its GPT-5.6 model series. The series includes three models: Sol, the flagship version; Terra, a mid-tier option; and Luna, the fastest and most affordable.

The review process followed an executive order issued on June 2, 2026, that established a voluntary framework allowing AI developers to submit models for government testing for up to 30 days before release. The Commerce Department's Center for AI Standards and Innovation conducted the review, focusing on national security risks and safety concerns.

Reports initially described the outcome as regulatory approval, but the White House pushed back on that framing. Officials clarified that private companies are not required to seek government permission to release AI models. The administration said the review was a voluntary safety check, not a licensing process.

For several weeks before the broad rollout, GPT-5.6 was available only to a small list of approximately 20 government-approved partners. OpenAI said it was uncomfortable with that arrangement becoming a long-term standard for the industry. The company confirmed plans for a broad rollout but did not announce a specific date.

U.S. lawmakers have separately opened probes into domestic companies' use of Chinese AI models, citing data security concerns. China has warned about risks from Anthropic's Claude Code, and Alibaba banned its employees from using Anthropic tools after allegations of a distillation attack, in which one AI model is used to train another without authorization.

Meta is also expanding its AI presence, entering the image model market and building its first major Canadian data center.