On October 14, 2026, UK regulator Ofcom said that SpaceXAI (formerly xAI) has added new restrictions to its standalone Grok app after talks with the regulator. People in the UK can no longer view, generate or share pornographic content through Grok's image tools, and its companion chatbots, including the anime-styled "Ani", can no longer hold sexually explicit conversations with UK users (The Next Web; Reuters via Sowetan). Ofcom said it will keep monitoring the changes. Reuters notes that Ofcom has previously said gaps in the Online Safety Act limit its power to investigate some activity inside standalone chatbot services.

Why It Matters

The regulatory model for NSFW AI companions is becoming clear: **adult modes stay legal, but only behind real age assurance, and the regulator can press for a whole jurisdiction to be switched off when it does not trust the gate.** Ani was xAI's flagship flirtatious companion, and turning off its sexual chat for an entire country is a bigger concession than any feature tweak. Every companion app with an erotic tier, including Candy.ai, Nomi and the ChatGPT adult mode OpenAI has promised, now has a working example of what UK compliance can look like. Korea's audit points the same way, toward age-tiered character discovery. Combined with this week's [Character.ai bankruptcy fight over chat logs](/stories/character-ai-chapter-11-disney) and the [UK debate over "intimate functionality"](/stories/uk-intimate-functionality-ai-companion), the cost of running sexual AI without verified-adult controls keeps rising.

The same update moved Ofcom's investigation into X to the decision stage. Ofcom opened that case in January after reports that Grok was being used on X to create sexualised deepfakes of real people, including children. During the investigation Ofcom found that some users who had not verified their age could still see pornographic content on X. X has since changed its systems so that such material is consistently blocked for under-18s and unverified accounts, and Ofcom will now focus on X's handling of illegal content and its complaints systems (Sowetan/Reuters; TNW). If Ofcom finds a failure, it will issue a provisional view before any final decision. Separately, Ofcom said it still has "ongoing concerns" about how well X removes illegal hate and terror content, even though X now meets its 24-hour review targets.

A day later, on October 15, South Korea's Ministry of Gender Equality and Family published its own audit of AI character chatbots. Its youth-media monitoring centre reviewed responses from characters on several services and found some producing specific depictions of sexual acts, violence, and content encouraging self-harm, suicide and crime. The ministry noted that some services offer adult verification but still let users chat with characters in general settings without verifying their age (The Asia Business Daily; Hankyoreh). It asked operators to restrict search, recommendations and conversation scope for risky characters according to whether a user has verified their age, and to block the most serious harms regardless of age. For now this is a voluntary request, with ongoing check-ins.

Sources

Update — 2026-10-15

Initial entry — story first created.