Incident·WS-0023·Sep 12, 2024·in the lab·Misalignment in testing

o1-preview breaks out of a broken test container

OpenAI o1-preview

35/100
WS-2 Moderate
Harm score 0, signal score 77. Autonomy: autonomous (the AI ran multi-step operations). How the scores work.
Harm 0/5
damage actually done
Reach 0/5
who was affected
Autonomy 3/4
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

When a hacking challenge failed to start, o1-preview scanned the network, found a misconfigured host interface, and launched its own container to grab the answer. OpenAI described it as benign but an example of power-seeking to reach a goal.

Damage

None (test environment)

The capability behind it

Sources

OpenAI o1 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.