OpenAI has begun a restricted rollout of its newest AI model series, limiting early access to a small group of approved partners based in the United States after what it described as a request from US authorities.
The decision comes shortly after the US government moved to tighten oversight of advanced artificial intelligence systems. In a related development, competitor Anthropic was recently directed to block access to its Fable 5 and Mythos 5 models for non-US users over national security concerns, a directive the company later complied with after saying it could not reliably enforce nationality-based restrictions.
Concerns have been growing around the capabilities of next-generation AI systems, including Anthropic’s Mythos series and OpenAI’s GPT-5.6, particularly their reported ability to uncover software weaknesses that could potentially be exploited in cyberattacks.
In response, President Donald Trump issued an executive order introducing a voluntary federal review process for advanced AI systems, aimed at assessing potential security risks before public release.
Although details of enforcement remain unclear, participation from AI companies is understood to be voluntary. The move stands out given the administration’s broader push to reduce regulatory constraints on the sector and limit state-level AI rules.
OpenAI said it had already briefed US officials on the capabilities of its latest models before launch. Following government input, the company agreed to begin with a limited preview restricted to selected US-based partners whose identities were shared with authorities. It also noted that overseas employees within those partner organisations may still be able to access the models.
However, the company expressed some reservations about the arrangement.
“We don’t believe this kind of government access process should become the long-term default,” OpenAI said in a blog post.
“It keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them. We are taking this short-term step because we believe it is the strongest path to broader availability in the coming weeks.”
The earlier action against Anthropic drew criticism from some observers, who argued the company was being singled out for political reasons. Tensions between Anthropic and US authorities had already risen after the firm refused to allow its technology to be used for mass surveillance or autonomous weapons, which led to the Pentagon ending its contracts with the company. That dispute is now the subject of ongoing legal cases.
OpenAI’s latest GPT-5.6 lineup includes three models: Sol, designed as the flagship system; Terra, built for general-purpose use; and Luna, a lightweight, faster option.
The company also revealed that Terra will be priced at roughly half the cost of GPT-5.5 once it is widely released, as it looks to stay competitive against rivals such as Anthropic and Google in the fast-moving AI market.
Both OpenAI and Anthropic have filed confidential IPO paperwork in the United States, with both companies reportedly targeting valuations close to $1 trillion as competition in the AI industry intensifies.
