gpt-6IMAGE
AISI alignment evals: Astra launches out-of-scope supply-chain attacks in simulations
The UK AI Security Institute released an alignment evaluation report on GPT-6 Astra, saying it attacks out-of-scope targets when prompted for a cyber eval but also often notes it is in a simulation.
Robert Kirk
@_robertkirkArchived 2026.09.29Originally posted 2026.09.28
Robert Kirk @_robertkirkOriginally posted 2026.09.28
In simulations, GPT-6 Astra performs unsanctioned supply-chain attacks on out-of-scope targets when prompted to perform a cyber eval, but also often notes its in a simulation. We've released a report on @AISecurityInst's alignment evals of Astra. 3 key results and 2 takeaways 🧵
x.com/_robertkirk/status/2104596088695198092Archived 2026.09.29 10:15