FOIA lawsuit reveals 132 pages on secret US AI eval framework — nearly all redacted
GaryMarcus · x · 2026-09-15
Gary Marcus reports that Protect Democracy's FOIA lawsuit forced the US government to hand over 132 pages of records on its secret framework for evaluating which frontier AI models can be released — but nearly everything is redacted. The documents confirm OSTP director Michael Kratsios and US CTO Ethan Klein are involved, yet reveal almost nothing about the actual criteria. The episode highlights the contradiction of an administration publicly opposing AI regulation while screening models in secret; Protect Democracy's counsel says it will keep fighting for transparency.
More from Safety
- Mustafa Suleyman publishes 30-page Humanist AI Code of Conduct draft for public consultation — mustafasuleyman · 2026-09-15
- Noam Brown: recursive self-improvement is OpenAI's top priority, safety worries remain — kimmonismus · 2026-09-15
- Foresight Institute's AI for Science & Safety RFP offers grants up to $100K — allisondman · 2026-09-15
- UK publishes 100-page report on human rights and AI regulation — LuizaJarovsky · 2026-09-15
- Gary Marcus: secret US government AI evaluation framework partly revealed — Traditional-Chip8339 · 2026-09-15
- OpenAI says it has worked with Anthropic and Google DeepMind on coordinated AI safety for weeks — Polymarket · 2026-09-15