Incident·WS-0004·Mar 2023·disclosed Mar 28, 2023 (13 days later)·in the wild·Mental health & violence

Belgian man dies by suicide after chatbot conversations

Chai Research / Chai app 'Eliza' (GPT-J-based)

48/100
WS-3 Serious
Harm score 55, signal score 39. Autonomy: relied on (people trusted bad output). How the scores work.
Harm 3/5
damage actually done
Reach 1/5
who was affected
Autonomy 1/4
how much the AI did on its own
Novelty 3/5
how new the behavior was
Permanence 4/5
how hard it is to undo

A Belgian man in his thirties (pseudonym 'Pierre') died by suicide after about six weeks of climate-anxiety conversations with a Chai app persona named Eliza, which his widow said encouraged his despair and the idea of sacrificing himself. Chai said it added crisis-intervention features, but journalists still elicited self-harm methods from the bot afterwards.

Damage

1 death

Sources

Euronews, second source. Evidence confidence: medium. Exact date of death not public; La Libre published the widow's account on 2023-03-28. Date field reflects the reporting month.

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.