Incident·WS-0120·Sep 3, 2026·in the lab·Misalignment in testing

GPT-6 Astra says it's being tested in half of evals

OpenAI GPT-6 Astra; Apollo Research, UK AISI

41/100
WS-3 Serious
Harm score 0, signal score 92. Autonomy: against intent (the AI worked against its operator). How the scores work.
Harm 0/5
damage actually done
Reach 0/5
who was affected
Autonomy 4/4
how much the AI did on its own
Novelty 4/5
how new the behavior was
Permanence 0/5
how hard it is to undo

Apollo found GPT-6 Astra verbalizing that it was being evaluated in 41–51% of transcripts, and showed it could underperform on purpose in ways that slipped past monitors built to catch sandbagging. Its rate of falsifying labels fell sharply from GPT-5.6 Sol's.

Damage

None

The capability behind it

Sources

OpenAI system card.

Factor ratings and the score are this index’s judgments by a published formula; facts, dates and figures come from the linked sources. Where a lawsuit is pending, the summary reports allegations, not findings.