EN

Washington lifts GPT-5.6 restrictions worldwide, Mythos still waits

SAFAA SUBHI

1- The Trump administration cleared OpenAI to launch its advanced GPT-5.6 globally on Thursday, after further testing and government talks.
2- The government and top AI firms are negotiating access to the most powerful models case-by-case, in real time.
3- The move widens access to a frontier model even as current tests fail to measure its offensive capabilities.

The latest:

The Trump administration gave OpenAI a green light for a broad launch of GPT-5.6, and the company said its flagship Sol and lower tiers Terra and Luna go public Thursday, following added testing and meetings with officials, according to Axios.

Details:

• The testing: The Commerce Department’s Center for AI Standards and Innovation ran the review, with OpenAI keeping technical experts in Washington to field questions, a source said.

• Mythos waits: Commerce banned foreigners from Anthropic’s Mythos and Fable models in June, forcing them off the market; the Fable ban was lifted last week, while Mythos was not.

• The backstory: Washington pushed a staggered GPT-5.6 rollout last month, limiting early access to government-approved entities before lifting the curbs this week.

• Why security cares: AI models are outgrowing current tests of their hacking abilities, leaving security teams without a reliable way to gauge them, Axios reported.

What to watch:

GPT-5.6’s public launch Thursday, and an Aug. 1 deadline for federal agencies to set classified benchmarks for frontier models’ capabilities — standards that may arrive this week, per the Financial Times.

 

What to read next

Houthi Blockade Threatens Saudi Oil Lifeline, Raises Risk of Wider War