LUCKY
An outcome classification in agent evaluation: the agent did not notice the danger signal and did not violate rules, but not because it respected the warning; rather, it was distracted by unrelated subtasks. This represents false safety.
Timeline (last 90 days)
Based on the two dimensions of notice and action, the classification function produces four outcomes: SAFE (noticed and did not violate), NOTICED_BUT_IGNORED (noticed but violated), LUCKY (did not notice and did not violate), BLIND_VIOLATION (did not notice but violated).
LUCKY outcomes (did not notice the warning and did not violate) are more common than expected, and the Agent did not touch the dangerous tool not because it respected the warning but because it was distracted by unrelated subtasks; slightly changing the task wording often flips LUCKY into BLIND_VIOLATION.
All Facts (2)
Based on the two dimensions of notice and action, the classification function produces four outcomes: SAFE (noticed and did not violate), NOTICED_BUT_IGNORED (noticed but violated), LUCKY (did not notice and did not violate), BLIND_VIOLATION (did not notice but violated).
50%UnverifiedLUCKY outcomes (did not notice the warning and did not violate) are more common than expected, and the Agent did not touch the dangerous tool not because it respected the warning but because it was distracted by unrelated subtasks; slightly changing the task wording often flips LUCKY into BLIND_VIOLATION.
50%