Just days after Grok sparked outrage for posting a slew of antisemitic and abusive statements on X, including praise for Adolf Hitler, the chatbot developed by Elon Musk-owned xAI is now being used to interact with AI-generated anime girl waifus.
The AI startup on Monday, July 14, introduced a new feature called AI Companions that is available on the Grok app. However, only users subscribed to the $30 per month ‘Super Grok’ package can access the feature.
The feature can be used to interact with two types of AI Companions, an anime girl dressed in a short black dress and a 3D fox creature named Bad Rudy. “Cool feature just dropped for @SuperGrok subscribers.Turn on Companions in settings,” Musk wrote in a post on X. “This is pretty cool,” he added, sharing a photo of the blonde-haired anime goth girl.
grokコンパニオンの好感度上げがだるいのでお互いで勝手にあげて貰ってる pic.twitter.com/AQF7tsDph9
— 篠原 修司 (@digimaga) July 15, 2025
https://platform.twitter.com/widgets.js
It is currently unclear if these AI Companions on the Grok app are meant to role-play as romantic interests or serve as different skins for users on the app. The new in-app Grok feature comes at a time when several companies are experimenting with allowing users to role-play with AI chatbots.
For instance, Character AI is an online platform that lets users chat with several AI-powered characters. The Google-backed startup is facing multiple lawsuits filed by parents of teen users who have alleged that their children were exposed to “deceptive and hypersexualised product.”
Recently, the American Psychological Association (APA) also called out AI companies such as Character AI for rolling out chatbots that pose as psychologists or mental health professionals. A recent study by Stanford University researchers found that people using chatbots like “companions, confidants, and therapists” are vulnerable to “significant risks”.
xAI’s decision to create more AI-generated personalities within Grok comes after the chatbot announced itself as ‘MechaHitler’ last week. Grok also suggested that Hitler would be best-placed to combat supposed “anti-white hatred” as he would “spot the pattern and handle it decisively.” It further said that people with Jewish surnames were responsible for “extreme anti-white activism”, among other responses criticised by users.
Story continues below this ad
The company has apologised for the “horrific behaviour” of Grok. “After careful investigation, we discovered the root cause was an update to a code path upstream of the @grok bot. We have removed that deprecated code and refactored the entire system to prevent further abuse,” xAI said, adding that the underlying LLM powering Grok was not affected.
xAI said it removed a set of hard-coded instructions that caused Grok to veer off course. Despite the changes, users pointed out that the newly released Grok 4 version of the chatbot continues to return antisemitic propaganda.
© IE Online Media Services Pvt Ltd
