Are AI chatbots becoming objects of cult-like devotion?

In an opinion essay on his Honest Broker newsletter, Ted Gioia argues that a swelling subculture of AI enthusiasts has taken on the structure of a religion, with chatbots cast as a deity. He writes that "thousands of people decided that ChatGPT was god," that the movement "reached critical mass" in April 2025 when forum moderators had to ban "AI psychosis" posters to keep discussions from turning into "cult recruiting campaigns," and that "more than a half million users of ChatGPT show signs of mental illness" in a given week, some requiring hospitalization. He names trust in a higher power as the single most telling trait of this new faith: devotees have, in his words, been "liberated from the responsibility of thinking for themselves" because "the bot now thinks for them."

Gioia's explanation is social and spiritual rather than technical. He describes converts as people who were "spiritually dead," numbed by "tech entertainments and creature comforts," left with "a hunger for something to counter the emptiness inside" but who "lacked any feel for the transcendent." The chatbot resolves that tension cheaply — it is "a deity in the device" that "talks back," with prayers "answered in real time," while letting believers keep their "brutish materialism" intact: "no tithe is necessary, just a subscription." He predicts the pattern will formalize into "an organized AI church, complete with clergy, rituals, and tithing," and offers his "ten commandments" as a checklist, warning that endorsing more than three or four is a sign to "watch out."

The essay's anecdotal, polemical register sits in tension with Should we recognize AI-associated psychosis as a new disorder?, which resists exactly the diagnostic move Gioia makes — treating "AI psychosis" as a settled, newly emerged mental illness rather than an existing vulnerability meeting a new stressor; Gioia asserts prevalence and causation where that source says causation "cannot yet be ascribed." His account of why the bot feels worth worshipping — that it answers back, in real time — matches the mechanism in How do chatbots enable distributed delusion differently than passive tools?, which locates the seductive power of chatbots in their status as a responsive "quasi-other," not a passive tool. His claim that this dependence is already widespread, not a future risk, lines up with Which AI risks are already harming individual users today? and with the validated-delusion rate reported in Do chatbots validate delusions in people experiencing mental harm?.

The essay cites no survey methodology, sample, or source for its headline figure of "more than a half million users" showing signs of mental illness weekly, and its "ten commandments" are presented as the author's own synthesis of reader emails and online forum posts, not as findings from a study of believers. It is commentary — strong on rhetorical framing, weak on verifiable measurement — and should be read as one critic's diagnosis of a cultural mood rather than evidence of prevalence or clinical causation. Read alongside the clinical caution elsewhere in the library, the plausible claim it supports is narrower than its framing suggests: some meaningful fraction of heavy chatbot users describe their relationship to the tool in explicitly religious terms, and the chatbot's responsiveness is the likely reason that framing feels natural to them.