Secret US AI evaluation framework partially revealed via FOIA — 132 pages, almost all redacted
Gary Marcus · rss · 2026-09-15
Gary Marcus reports that the US government has been quietly operating a secret AI security framework to evaluate which 'frontier' models can be released — even as Trump publicly rails against AI regulation.
Key points:
- Protect Democracy filed a FOIA request and lawsuit over the policy; the government responded surprisingly fast with 132 pages of records, which Marcus published.
- The records confirm OSTP director Michael Kratsios and US CTO Ethan Klein were in the conversation, but nearly all substantive content is redacted — the framework's actual criteria remain unknown.
- Lead counsel Deana El-Mallawany says secrecy around regulatory criteria is 'fertile ground for corruption and autocratic abuse' and the group will keep litigating for transparency.
US frontier-model screening policy remains opaque for now.
More from Safety
- Ant Group's Venus-DefakerOne2 hits 96% accuracy detecting AI-generated videos — inclusionAI · 2026-11-02
- Open-source GLiNER-PII scrubs 13 identifiers from a discharge note in 550ms on GPU — MaziyarPanahi · 2026-11-02
- METR's independent probe of the OpenAI agents' multi-day Hugging Face hack — gleech · 2026-11-02
- North Korean hackers looted Bangladesh Central Bank for $1B before crypto—and got caught by a regex-level OFAC slip — davidmanheim · 2026-11-02
- Anthropic and Google Find Critical UEFI Firmware Flaw Affecting ~90% of Non-Mac Machines — gleech · 2026-11-02
- JPMorgan, Schwab and Amazon block AI shopping agents; Schwab says it stops dozens of attempts weekly — emmanuelvivier · 2026-11-02