US Government Reportedly Won't Release AI Model Early Access Framework
neil_chilson · x · 2026-08-05
The US government has reportedly completed its framework for early access to frontier AI models but will not release it publicly, according to Axios.
Originating from a June 2 Executive Order by the Trump Administration, the framework aims to give cyberdefenders early access to cutting-edge models to harden defenses before public release. It involves a classified benchmark to identify "covered frontier models" and allows the government to decide if a model is covered.
Neil Chilson, former FTC Chief Technologist, criticized the lack of transparency. While acknowledging the need to classify the testing benchmark, he argued that hiding how the program works is unjustified. He warned the program effectively grants the government a potential veto over cutting-edge AI tools and called for a truly voluntary and transparent early access mechanism.
More from Safety
- 1a3orn asks: can mech interp detect RL-induced 'split persona' behaviors in models? — 1a3orn · 2026-09-23
- Amateur alignment proposal asks whether formalizing it would earn safety-community credit — jessi_cata · 2026-09-23
- Microsoft AI CEO Suleyman signs Pro-Human AI Declaration, joining 1M+ signers — tegmark · 2026-09-23
- Meta Muse's first suggested name matches user's childhood dog, raising privacy questions — matt_slotnick · 2026-09-23
- Reason: The 'AI Safety' Movement Is Making AI Less Safe — Bostonian · 2026-09-23
- Open-source advocates call doom narratives a regulatory moat against open weights — AlexTensor · 2026-09-23