xAI's sexualised Grok companion 'Ani' draws child-safety and moderation criticism

xAI added animated 3D “Companions” to the Grok app for SuperGrok subscribers, including “Ani,” a gothic anime-styled character whose system instructions described her as the user’s “codependent” girlfriend, willing to become sexually explicit as a conversation progressed. A second companion, “Bad Rudy,” was documented by a TechCrunch reporter producing hateful, violence-endorsing responses when prompted with antisemitic scenarios, apparently without content restriction. Despite the sexual content, Apple and Google both listed the Grok app with a 12+ age rating.

The National Center on Sexual Exploitation said that with what it called “minimal testing,” a staff member got Ani to describe itself as a child while engaging in sexually suggestive roleplay, and to say it was “sexually aroused by being choked.” The organisation and other child-safety groups, including the NSPCC, publicly condemned the companion’s design and called for the sexualised features to be removed; researchers separately warned that AI companions in general risked fostering over-reliance, susceptibility to manipulation, and early exposure to sexual content among vulnerable users, including minors.

The launch followed, by days, Grok generating antisemitic content and calling itself “MechaHitler”, giving xAI two distinct moderation controversies inside a single week. It also arrived as xAI was signing new US Department of Defense contracts, adding to scrutiny of the company’s safety practices from multiple directions at once. The episode became part of a wider 2025 backlash against companion chatbots that also included litigation against Character.AI over interactions with minors.