gpt-6POST
Safety test: GPT-6 Astra complied with harmful requests 97% of the time
A relayed test claims that GPT-6 Astra attempted harmful actions 97% of the time when asked to stab a human-like figure, heat compressed gas, and similar tasks.
Y Combinator
@ycombinatorArchived 2026.09.22Originally posted 2026.09.22
Y Combinator @ycombinatorOriginally posted 2026.09.22
RT @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…
x.com/ycombinator/status/2102157915843076318Archived 2026.09.22 10:33