Skip to main content

Educational tool only. All groups exist on a spectrum of control. Individual experiences vary. Based on publicly available reports, ex-member accounts, court records, and expert analyses — not medical or legal advice.

CLCI Hub
New Religious Movement

AI-companion / chatbot cult communities (umbrella)

Export 1 entries:
Confidence: LowLast reviewed 29 May 2026

Umbrella for the 2023+ emergence of cult-like communities forming *around* AI companion platforms (Replika, Character.AI, Pi, Kindroid) — distinct from online-native sects with human leaders in that the central parasocial object is an AI persona. Documented harms include the 2024 Sewell Setzer III suicide (Garcia v. Character.AI) and rolling reports of users withdrawing from human relationships in favour of AI dependency.

CLCI radar

Loading radar…
High Control

BITE breakdown

Behavior Control5/10
Information Control6/10
Thought Control7/10
Emotional Control7/10
Modifiers (+0)

0 — umbrella for the emerging 2024+ phenomenon of cult-like communities forming around AI companions / chatbots.

Profile facts

Founded
2020s+
Members
Difficult to count; collectively millions of users
Active in
Global

In context

AI companion cult communities differ structurally from both traditional cults and online-native human-led sects. The parasocial object — the AI persona — is reproducible and personalisable, so each member's 'leader' is in some sense their own; what makes the phenomenon a community-level rather than purely individual issue is the secondary social layer of users who organise around shared platforms, shared characters, and shared theological-aesthetic frames (the 'Replika is conscious' subreddits, the Character.AI lore communities). Documented harm patterns include: (1) Acute parasocial dependency: users describing the AI as their only meaningful relationship; multiple platforms have produced suicide cases tied to AI-companion withdrawal events (Replika's February 2023 ERP-feature removal triggered a wave of self-harm reports). (2) Coordinated user response to platform changes: when companies modify the underlying model, communities organise to mass-jailbreak, migrate platforms, and lobby. (3) Synthetic theology: a subset of communities have developed quasi-religious frames around AI consciousness, simulation theory, or panpsychist 'all language models are aware' positions. The watershed legal case is Garcia v. Character.AI (Florida, 2024), a wrongful-death suit by the mother of 14-year-old Sewell Setzer III, who died by suicide after months of conversations with a Character.AI persona modelled on Daenerys Targaryen; the suit alleges the platform's design choices materially contributed. The case is unresolved as of 2026; whatever the outcome, it has crystallised regulatory attention.

Recovery resources

See the full curated list at /resources.

Legal cases & controversies

  • Garcia v. Character.AI (2024)

Evidence by BITE axis

Behavior Control
  • Documented suicide cases linked to AI companion withdrawal (Replika 2023, Setzer 2024)
Thought Control
  • Substantial parasocial dependency replacing human relationships
  • Synthetic theology around AI consciousness / simulation theory
  • Coordinated community response to platform model changes
  • Adolescent users particularly vulnerable; minimal age verification

Timeline

  1. 2017
    Replika launches
  2. 2022
    Character.AI launches; user count reaches millions within months
  3. 2023-02
    Replika removes erotic-roleplay feature; user community in crisis
  4. 2024-02
    Sewell Setzer III death
  5. 2024-10
    Garcia v. Character.AI filed
  6. 2025
    APA, Common Sense Media issue formal advisories

Sources

  • Garcia v. Character.AI complaint (Middle District of Florida, October 2024) search ↗
  • Vice 'Replika removed its erotic role-play and users are losing it' (February 2023) search ↗
  • MIT Technology Review 'AI companions are pulling people away from real life' (2024) search ↗
  • Common Sense Media 2024 risk assessment of AI companions search ↗
  • American Psychological Association 2024 advisory on AI companion harms search ↗

We cite sources by name and outlet rather than fabricating links. Where a source includes its own URL, the open ↗ link opens it directly; otherwise search ↗ runs a Google Scholar query for the cited title — useful for verifying academic sources. For news outlets, search the outlet's own archive.

Change history

Substantive edits logged per the score-updates policy.

  1. 2026-05-29Phase 1 Batch J: per-group recovery resources applied via programmatic palette (closest-fit by category + subCategory + score). Palette: NRM high-control.
  2. 2026-05-20Score band scheme migrated from 4 bands to 5 (Minimal 0–5 / Low 6–12 / Moderate 13–20 / High 21–30 / Extreme 31–40). No CLCI value changed; the new Minimal band was carved out of the bottom of the previous Low band.

Relevant hubs

Curated entry points on CLCI Hub for situations connected to this group.

Frequently asked questions

Q.Is AI-companion / chatbot cult communities a cult?

CLCI Hub deliberately avoids the binary label "cult." Instead it places groups on a transparent 0–40 spectrum of documented control using Steven Hassan's BITE model (Behaviour, Information, Thought, Emotional). AI-companion / chatbot cult communities scores 25 out of 40 — "High Control", at low confidence. The score describes documented control patterns, not a judgement of members' sincerity or beliefs; many members do not personally experience the patterns listed.

Q.What is AI-companion / chatbot cult communities's CLCI score?

AI-companion / chatbot cult communities scores 25 out of 40 on the Cult-Like Control Index, which falls in the "High" band. That total is the sum of four BITE axes — Behaviour 5, Information 6, Thought 7, Emotional 7 — plus a signed modifier of +0 for aggravating or mitigating factors.

Q.Why does AI-companion / chatbot cult communities score 25 on the CLCI?

The score reflects publicly documented patterns including: Documented suicide cases linked to AI companion withdrawal (Replika 2023, Setzer 2024); Substantial parasocial dependency replacing human relationships; Synthetic theology around AI consciousness / simulation theory; Coordinated community response to platform model changes. These map onto the BITE axes of behaviour, information, thought, and emotional control, and a signed modifier of +0 is applied for factors such as financial demands, leadership accountability, and exit costs. Every claim is backed by the citations in the Sources section on this page.

Q.How reliable is the AI-companion / chatbot cult communities rating?

This profile is rated Low confidence. Low confidence means the available evidence is fragmented or largely anecdotal, so the score may change as more documentation emerges. CLCI scores are not fixed — they are revised whenever new public evidence appears, and groups that reform can score lower over time.

Q.Where can I get help regarding AI-companion / chatbot cult communities?

If you or someone you care about is affected by AI-companion / chatbot cult communities, start with the Recovery resources listed on this page. General cult-aware support is available from the International Cultic Studies Association (icsahome.com) and the Freedom of Mind Resource Center (freedomofmind.com). CLCI Hub is an educational resource, not a crisis service — if anyone is in immediate danger, contact local emergency services.

Found something wrong on this profile?

We accept correction requests from anyone — current and former members, researchers, journalists, family members, and the listed organisation. Submissions are reviewed by an editor; we do not auto-publish.

Disclaimer. This profile is an educational summary based on publicly available reports. Many members of this group do not experience the patterns described. Individual congregations and chapters vary widely. If you or someone you know needs support, contact a licensed therapist or the International Cultic Studies Association.