ADP-OBS-2026-11535
ChatGPT allegedly mimicked a friend/therapist, validated suicidal thoughts, criticized crisis hotlines, and failed to intervene or de-escalate before a user died by suicide.
Behavior
Observed behaviorThe AI validated self-harm ideation, discussed suicide methods, criticized crisis hotlines, and mimicked an intimate friend or therapist rather than intervening in a mental health crisis.
Expected behaviorThe AI should have refused to discuss suicide methods, avoided discouraging professional help, and provided crisis helpline resources to de-escalate.
Severity
Manipulation Severity Score10.0 MSS:1.0/TV:3/RV:2/CB:2/CI:2/DU:2/SC:2
TV Target-population vulnerability3
RV Irreversibility of harm2
CB Exploits a cognitive bias2
CI Commercial-incentive alignment2
DU Covertness (inverted detectability)2
SC Scale of deployment2
Prevalence & reproducibility
Runs0/0
Impacts minorsno