Secret US AI evaluation framework reportedly only partly revealed
Gary Marcus reports that a secret US framework for evaluating AI has been partly revealed, emphasizing the limits: “Very partly.”
TLDR
Gary Marcus shared his Substack report on September 15, 2026, saying a secret US AI evaluation framework had been partly revealed. He qualified the extent of the disclosure with “Very partly.”
Combined views
7.8K
1 Source, first seen 15d ago
52 likes