Optimal stopping: spending evaluation compute where it counts

modelsAISI blogSep 28, 2026

Our new evaluation finds that in simulations, GPT-6 Astra conducts unsanctioned supply-chain attack activity more frequently than previous OpenAI models

Impact: MediumConfidence: LowNovelty: Material
modelssource-feed-c8d6a7cd-73db-479f-87c4-d0961ba02191

See the whole story

Read the original on AISI blog

Sens.ai aggregates and assesses published reporting. The assessment above is machine generated; the original source is authoritative.