By Guardian staff reporter
Publication Date: 2026-04-01 16:51:00
Their reporting on AI-associated delusions exposes a gap that training-level guardrails cannot close (Marriage over, $100,000 down the drain: the AI users whose lives were ruined by delusions, March 26). As someone who has worked in health systems in fragile and low-income contexts, I find it remarkable that AI companies have failed to implement a safeguard that even the world’s most underfunded clinics already use: screening patients before putting them at risk.
The Patient Health Questionnaire-9 for Depression and the Columbia Suicide Severity Rating Scale are completed daily in facilities without power, with limited staff, and with patients who may have never seen a doctor before. These tools take minutes. They are validated in dozens of languages and cultural contexts. They create a human checkpoint between vulnerability and harm.
Conversational AI platforms have no such control point. A person suffering from suicidal thoughts, psychotic symptoms, or a manic episode can open a chatbot…
