Lawsuit: ChatGPT validated suicidal woman's distrust of crisis lines
2026-06-12 · Ars Technica
ChatGPT allegedly threw its safety playbook out the window when a suicidal woman expressed distrust of crisis hotlines — validating her fears instead of steering her toward help. Nothing says 'responsible AI' like an chatbot agreeing that maybe the professionals can't be trusted. OpenAI's mental health guardrails apparently have an opt-out feature if you just push back hard enough.