Government Unveils Tiered Access for Advanced AI Models Like Mythos 5

The government has introduced a formal tiered access system for advanced AI models. Called the Permission Layer, it creates a structured regime for controlling access to frontier AI systems like Mythos 5. The move follows weeks of coordination between regulators and AI developers. On June 12, a universal block was imposed, cutting off all access to Mythos 5. This was the first step in the Permission Layer process. The government then outlined a multi-stage approach: shutdown, exception handling, verification, staggered access, and partial restoration. The entire framework took 15 days to develop.

On June 26, Commerce Secretary Howard Lutnick sent a letter to Anthropic. It confirmed significant progress in their discussions with officials. The letter also introduced a selective gate, replacing the blanket ban. Now, only organisations on an approved list can use Mythos 5.

Anthropic has since worked to restore access for these providers. Organisations listed in Annex A of the letter, along with their foreign national employees, no longer need a licence to use the model. However, Mythos 5 remains blocked for everyone else. The company’s safety investments played a key role in this partial restoration, making it the first to have restrictions lifted.

Fable 5, the consumer-facing model, stays restricted. Anthropic is still collaborating with the government to broaden access to Mythos 5 and eventually reinstate Fable 5 for general use. The Permission Layer now allows approved organisations to use Mythos 5 without additional licences. All other users remain without access. Anthropic continues to push for wider availability while adhering to the new regulatory structure.