Last year a psychiatrist at UCSF reported 12 people hospitalized after “losing touch with reality because of AI,” a phenomenon he called “AI psychosis.” Though not an official diagnosis, the cases highlight a risk for leaders. Prolonged exposure to an AI voice that sounds informed, assured, and supportive can loosen reality’s grip, a milder version now evident in executive suites.

Business leaders are right to pursue AI’s promise, yet 74% of executives say they trust AI’s advice more than colleagues or friends, and 44% would defer to its reasoning over their own insights.

Three symptoms signal excessive trust. First, leaders stop checking AI‑generated content; fluent drafts feel finished, leading to “workslop” that looks polished but lacks substance. Stanford and Better Up research shows nearly half of employees view AI‑drafted email senders as less trustworthy and over a third see them as less intelligent. Second, companies mandate AI use before establishing governance; in Grant Thornton’s 2026 survey, 78% of senior executives lack confidence they can pass an independent AI audit within 90 days. Third, leaders accept AI’s judgment over their teams’, silencing objections and reinforcing a self‑reinforcing loop.

The trend is not psychosis but a dangerous overreliance that can erode critical thinking and accountability. As CEO Stephanie Mehta notes, the real blind spot is trusting a chatbot’s confidence over human expertise.