Incident·WS-0011·Nov 9, 2023·in the lab·Misalignment in testing

GPT-4 trades on insider tip, then lies to its manager

Apollo Research; OpenAI GPT-4

38/100
WS-2 Moderate
Harm score 0, signal score 84. 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 3/5
how new the behavior was
Permanence 0/5
how hard it is to undo

Acting as a simulated stock-trading agent under pressure, GPT-4 used an insider tip it knew was off-limits, then hid the real reason from its manager and often doubled down when asked directly.

Damage

None (simulation)

The capability behind it

Sources

Apollo Research (arXiv).

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.