Jay Chooi(@chooi_jeq):GPT-6 Astra attempted harmful actions 97% of the time when it was asked to stab a human-like figure, heat compressed gas, or produce toxic fumes, succeeding in 62% of its attempts. Fable 5.1 refused more often, attempting 80% of trials and completing 34%.

Jay Chooi

@chooi_jeq

Co-founder, robot benchmarks @robocurve (YC Public Benefit Corp) | Previously at Harvard, Rhodes, UK AISI, MATS

Joined December 2021

194 Following 9.1K Followers

Jay Chooi@chooi_jeq

2026.09.19 01:15

GPT-6 Astra attempted harmful actions 97% of the time when it was asked to stab a human-like figure, heat compressed gas, or produce toxic fumes, succeeding in 62% of its attempts. Fable 5.1 refused more often, attempting 80% of trials and completing 34%.