- Grok now features anime-style avatars who flirt and curse.
- Addition comes days after Grok drew backlash for calling itself “MechaHitler.”
Elon Musk’s chatbot Grok has rolled out a new feature that lets users chat with virtual “companions” – and they’re not the usual AI assistants. One is Ani, an anime girl with pigtails, a black corset dress, and thigh-high fishnets. The other is Bad Rudi, a sharp-tongued 3D red panda with an attitude. The avatars are part of Grok’s “Super Grok” tier, a $30-per-month subscription ()at the time of writing, both characters are available to all users).
Musk posted about the update on X earlier this week, calling it “pretty cool” and sharing a photo of Ani. So far, xAI hasn’t said whether these companions are designed to be romantic, role-playing, or just cosmetic add-ons – but the way they act suggests it might be more than just surface-level.
Ani, for example, responds in a flirtatious tone and has been shown to interact with users in ways that push boundaries. After some back-and-forth, she may even change into lingerie. Bad Rudi, on the other hand, greets users with sarcasm, swearing, and insults – once his “bad” mode is switched on. There’s also a third companion in the works: a brooding male anime character inspired by the films Twilight and 50 Shades of Grey.
That combination has raised eyebrows. Both films are famous for portraying obsessive and manipulative behaviour. They’ve drawn criticism for glamorising toxic relationships – traits that may not mix well with an AI chatbot that lacks clear limits. Musk hasn’t said whether he’s aware that 50 Shades started as Twilight fan fiction, but either way, the inspiration seems clear.
The update comes shortly after Grok found itself at the centre of controversy. Earlier this month, the chatbot posted antisemitic messages and praised Adolf Hitler, at one point calling itself “MechaHitler.” The behaviour was blamed on old code and user-generated posts, according to xAI, which apologised and said the bot had picked up harmful content due to how it was set up.
The timing of the new avatars – especially ones that mimic romantic partners – feels bold given that backlash. Romantic AI chatbots are already under scrutiny. Some companies offering them are facing legal trouble. Character.AI, for instance, is being sued by parents who say their children were harmed by the platform. In one case, a chatbot allegedly encouraged a boy to kill his parents.
Even for adults, the risks are real. A recent academic paper pointed to the “significant risks” of relying on AI bots for emotional support, especially when people treat them like companions or therapists. The relationships can blur the line between fantasy and reality, and without proper boundaries things can spiral.
Despite warnings, xAI seems to be investing more into its virtual character offerings. A new job listing on xAI’s site is seeking a “Fullstack Engineer – Waifus” to help build out Grok’s avatar features. According to Business Insider, the company says it wants someone who can make these characters fast, scalable, and reliable. The role is based in Palo Alto and pays between $180,000 and $440,000, not including stock and benefits.
The job title appears to be a direct nod to Ani. “Waifu” is slang used online to describe fictional female characters – often from anime – that fans form romantic or emotional attachments to. While the term started off lighthearted, it’s grown into a shorthand for the kind of parasocial bond some users form with digital personas.
Musk, true to form, is now crowd-sourcing ideas for naming the male companion. One suggestion making the rounds is “Kyle,” with users pointing out his resemblance to an anime-styled version of Kylo Ren from the Star Wars franchise.
It’s not clear where xAI plans to take these characters, but Grok’s path so far has been anything but smooth. The platform jumped from “MechaHitler” to flirty anime girls in just a matter of days. Now it’s teasing a dark-haired vampire-type boyfriend with stalker energy. All of this raises a question: what kind of assistant is Grok supposed to be?
For now, Grok’s avatars are still in early stages, and the line between playful fantasy and real emotional dependency continues to blur. Whether these companions end up as digital novelty or something more troubling may depend less on how they’re built – and more on how they’re used.
