my bet is they're classifying the benchmarking process to hide the fact that they're not going to be able to agree to a regulatory threshold better than 10^26 flop
10:13 AM · Jun 2, 2026 · 15.2K Views
Dean W. Ball argues regulators might classify AI benchmarks to hide an inability to agree on thresholds stricter than 10^26 FLOPs
This highlights the debate between independent evaluations and self-reporting.
--0--
Sentiment
Sentiment building, check back later.
Cluster Engagement
-
Views
-
Comments
-
Reposts
-
Bookmarks
Expand data
Posts from X
Most Activity
Most ActivityTimeline
VIEWS2.9KLIKES36
Dean W. Ball@deanwball
Total @ohlennart victory
Dean W. Ball@deanwball
my bet is they're classifying the benchmarking process to hide the fact that they're not going to be able to agree to a regulatory threshold better than 10^26 flop
10hViews 2.9KLikes 36Bookmarks 0