OpenAI Restricts Preview of New AI Models to US Partners Following Government Request

OpenAI has launched a limited preview of its latest AI models exclusively for select US-based partners, following a request from the US government amid growing national security concerns over advanced artificial intelligence.

OpenAI on Friday announced a restricted preview of its newest and most powerful AI model family, making the models available only to a select group of trusted organisations in the United States.

The move comes just weeks after the US government imposed similar restrictions on rival AI company Anthropic, signalling increased federal oversight of cutting-edge artificial intelligence technologies.

US-only rollout for trusted partners

According to OpenAI, the company briefed US government officials on the capabilities of its latest AI models before their release.

At the government’s request, the initial rollout has been limited to selected US-based partner organisations whose identities have been shared with federal authorities.

OpenAI clarified that while participating organisations are headquartered in the United States, employees located overseas will also be able to use the new AI models through their respective companies.

New AI model family introduced

The latest release includes three new AI models, each designed for different use cases.

• Sol – OpenAI’s flagship model for advanced reasoning and complex tasks.
• Terra – A general-purpose AI model built for everyday productivity.
• Luna – A lightweight, faster and lower-cost model intended for high-volume applications.

The company also said Terra will eventually cost around 50% less than GPT-5.5, as OpenAI continues competing with Google and Anthropic in the rapidly growing AI market.

Restrictions follow White House AI security measures

The limited rollout comes after President Donald Trump signed an executive order earlier this month introducing a voluntary federal review process for advanced AI systems before they are released publicly.

The administration has argued that increasingly capable AI models should undergo national security assessments to reduce the risk of misuse.

Anthropic faced similar restrictions

OpenAI’s decision follows recent government action involving AI startup Anthropic.

Earlier this month, the White House directed Anthropic to restrict foreign access to its advanced Fable 5 and Mythos 5 models over national security concerns.

Anthropic temporarily suspended access to both systems, saying it could not reliably enforce restrictions based on users’ nationality.

Government eases restrictions on Anthropic

In a separate development, the US government has now lifted restrictions on Anthropic’s Claude Mythos 5 model.

The decision allows the company to provide access to more than 100 US organisations, including government agencies, research institutions and major private-sector companies.

Officials said the move followed additional discussions regarding security safeguards after concerns were raised that highly capable AI models could potentially be exploited for malicious purposes.

AI regulation continues to evolve

The latest actions involving both OpenAI and Anthropic highlight the US government’s growing focus on balancing rapid AI innovation with national security.

As more powerful AI systems emerge, companies developing frontier models are expected to face closer government scrutiny before making their technologies broadly available.