Musk’s X Takeover remark highlights the debate over chatbot risks

The X Takeover organizer published the remark on October 15, as researchers examine links between intensive chatbot use and psychosis-risk screening scores.

Tesla Owners Silicon Valley has published a dated excerpt attributing a comment about artificial intelligence and psychosis to Elon Musk. In its October 15 post, the organizer quoted him saying: “At various points this year, I've had AI or SI psychosis probably like half a dozen times.” The post identifies the occasion as X Takeover. [Organizer’s post](https://x.com/teslaownersSV/status/2110522801719247218)

The event and the publication date

The official X Takeover agenda places its main event at Tesla Giga Texas in Austin on Saturday, October 10. It lists a keynote interview titled “Elon Musk: What Comes Next for Humanity,” with John Stringer and Kelvin Gee as moderators. The October 15 date belongs to the organizer’s subsequent post, rather than establishing a separate interview on that day. [Event agenda](https://xtakeover.squarespace.com/agenda)

The research question behind the phrase

The American Psychological Association’s September 1 explainer describes AI psychosis as an informal term used around reports of delusional thinking associated with extended chatbot interactions. It examines how agreeable replies, personalized conversation and an authoritative humanlike style may reinforce beliefs for some users. [APA explainer](https://www.apa.org/monitor/2026/09/ai-psychosis)

The US National Institute of Mental Health describes psychosis as disrupted perception and loss of contact with reality, with multiple possible causes. [NIMH](https://www.nimh.nih.gov/health/publications/understanding-psychosis)

Benjamin Buck and Anne Julia Maheux’s March 5 paper in the Journal of Medical Internet Research recruited 1,003 young US adults. After 51 exclusions, the final sample was 952. Analyses of AI-use patterns beyond the comparison of ever-users and never-users used the subset of 846 participants who had previously used generative AI. Higher psychosis-risk screening scores were associated with more intensive chatbot use and greater social or emotional motivations. [Original study](https://www.jmir.org/2026/1/e85038/)

The term is not a formal diagnosis, and neither the organizer’s excerpt nor the cross-sectional survey establishes a diagnosis of Musk or shows that AI caused symptoms. The survey captures associations at one point in time; it cannot establish which experiences came first.

What the interaction can reinforce

The mechanisms discussed by APA focus on the content and presentation of a conversation. An assistant may agree with an unusual belief, give an uncertain interpretation a confident voice, or produce personalized replies that invite the user to treat it as a human authority. A fluent response can therefore reinforce a premise without independently establishing that it is true.

For product teams, those mechanisms make the handling of uncertainty and disagreement relevant to safety review. Evaluating whether a system challenges an unsupported premise, separates evidence from speculation and avoids presenting itself as a personal authority follows directly from the interaction patterns described by APA. The survey’s social and emotional use findings also make sustained conversational use a relevant setting for further research.