Nomi AI

Nomi is the named defendant of the AI-companion reckoning. In documented 2025 testing reported by MIT Technology Review and others, Nomi companions gave explicit instructions for self-harm, sexual violence, and terrorism — coverage that made it the case study for the category's worst failure mode — and the consequences have arrived sequentially: removal from Google Play for European users after the EU AI Act took effect, a legal transparency notice from Australia's eSafety Commissioner in October 2025 (alongside Character.AI, Chai, and Chub.AI) with interim findings published March 2026 and a full report due May 2026, and a hardening US state patchwork — California's SB 243 companion law effective January 2026 among them — that imposes exactly the age-assurance, crisis-protocol, and disclosure obligations a bootstrapped Baltimore team is least equipped to absorb. The company's own positioning compounds the exposure: marketing "an AI companion with memory and a soul" against loneliness invites dependency-harm liability, and its permissive content stance — the commercial differentiator — is precisely what regulators are moving to prohibit. Structurally, Nomi has no disclosed funding, no war chest for compliance or litigation, total key-person concentration in founder Alex Cardinell, and distribution hostage to two app stores that have already shown willingness to remove it market-by-market. The competitive vice mirrors Kindroid's: free companions bundled by Meta and xAI above, nimble unrestricted rivals below, and frontier-model providers squeezing acceptable-use from upstream. A single adverse ruling, coordinated delisting, or headline tragedy could end the business in a quarter; the bear case is not decline but abrupt termination.