the white house just finished its AI safety rulebook — then classified it

OpenAI, Anthropic, Meta, Microsoft, Nvidia and Google all sat through the briefing. none of us get to read what it actually says. Aliteq Lena Fischer · AI & Local Compute Editor

Executive order signed June 2, 2026 gave the administration 60 days — an August 1 deadline — to build a frontier-model evaluation framework

The framework lets the government request early access to a frontier model for up to 30 days before public release, to test 'advanced cyber capabilities'

It explicitly excludes mandatory licensing or preclearance requirements — it's framed as voluntary

OpenAI, Anthropic, Meta, Microsoft, Nvidia, Google, Cloudflare, JPMorgan Chase and smaller firms were briefed August 4

The government has not published the framework's contents, benchmarks, or the reasoning behind keeping it classified

Chris McGuire of the Council on Foreign Relations put it bluntly: 'We can't have secret, voluntary rules to regulate the most important tech in the world.' Nobody from the administration has offered…

Read the full story

Aliteq the white house just finished its AI safety rulebook — then classified it Read the full story on Aliteq