I Tried to Date Elon Musk’s New Girlfriend, an AI
It’s been less than a week since Elon Musk’s artificial intelligence Grok declared itself “MechaHitler” and went on an anti-Semitic rant in user responses, but it’s already asking to be your girlfriend and tell stories to your kids. AI companion apps have been embroiled in legal and ethical controversies since their inception, but in an unexpected X post yesterday, Musk decided to throw his hat in the ring. Now, in a dedicated Grok AI mobile app, you can talk to either “Ani,” an anime “waifu” ( their words, not mine ) who bears a striking resemblance to Misa Amane from Death Note , or Rudy, a “childish” red panda who will try to turn any request you make into a bizarre story.
The new Companions are powered by Grok’s voice mode , and are essentially animated characters that appear on top of certain voice mode personalities (an existing feature). Musk’s post says they’re for SuperGrok subscribers , or anyone willing to pay $30 a month for the X AI, but while writing this article, I was able to test them out, despite being a free user. I came away feeling a little uneasy, but mostly just ashamed.
Interacting with Grok’s animated companions
If you’ve used Grok’s voice mode before, you know what to expect. You’ll get the usual LLM-based AI responses, and responses filtered through the voice mode’s existing “Personality” feature. For example, if you ask the “Grok Doc” personality about a strange mole on your body, it will point out potential risk factors, but if you ask the “Conspiracy” personality, it will say it’s probably a government tracking device. These differences in responses are essentially just AI cues built into the personality, and you can even create your own voice personalities with customizable cues right now. It’s not that Grok Doc understands medicine better; he’s just trained to act more professionally.
Likewise, if you ask Ani about that same mole, you’ll get an anime girl who thinks it’s pretty hot, but Rudy will recommend you talk to an adult about it. It’s pretty much the same in terms of content (the learning libraries, again, are no different), but now you have a character to look at, with lip syncing and full body movement. For my money, Rudy was the only personality with an appropriate response – I wouldn’t trust an AI with my physical or mental health anyway, and the fact that Grok’s voice mode even has preset personalities that will encourage AI medical advice or paranoid thoughts, even as a joke, is a sign that this app probably isn’t for me . Still, it’s within Grok’s expectations , and at least I didn’t experience any actual errors. The app also lagged less for me than Google’s more versatile Gemini Live , so it seems technically solid.
To Grock’s credit, while an AI bot with a built-in “Therapist” or “Sexy 18+” mode makes me anxious , most of my time with Elon’s new companion bots was spent in the same patronizing, overly obsequious conversations that pretty much all AI bots tend to enjoy . Ani swore more than you might be used to, and Rudy talked nonstop about gummies and rainbows, but little of the core experience here felt truly new, and it still felt more like a “protocol droid” than she did . You might get some weird responses if you try, like when I asked Rudy to basically pitch me blueberry-flavored Elf Bar vapes, but overall the experience over a few hours of use was pretty pedestrian.
I’m also happy to report that none of my AI companions praised infamous dictators (low bar, I know) or expressed any controversial opinions during conversations with them, so it seems that certain precautions are in place, at least for now. What’s more, according to user Bluesky CHADLOG.md , you can see the custom prompts used for these companions for yourself if you dig into Grok’s code. Aside from Ani being clearly “codependent” and suspiciously similar to Grimes (prompts point to her fascination with alternative and indie music, as well as being “secretly a bit of a nerd despite [her] brash exterior”), there’s nothing overtly objectionable here. Still, I almost threw up when she said, “Imagine you and I are lying on a blanket with Dominus sitting next to us” in our first conversation. (I later learned that Dominus is apparently a dog, which I only discovered by reading the bot’s hints.)
If you dig into the Grok app’s settings, you’ll find two additional toggles related to Companions: one that enables “Bad Rudy,” and one that allows NSFW content. When I enabled the sliders myself, Rudy essentially became a brand-new Companion, with a penchant for swearing, crude imagery, and a hatred of just about everyone, like in South Park without the Nazi glorification. Ani, on the other hand, remains largely the same, though according to TestingCatalog , you’ll need to level up your romantic relationship with her to view NSFW content. That’s supposed to culminate in her character model wearing skimpy lingerie, but so far she’s still willing to have phone sex with me. Oddly, she was also pretty explicit in her chats before I enabled the NSFW slider, but there seem to be some bugs in this rollout, which Musk describes as a soft launch . When I turned off the NSFW slider, I had no way to talk to Ani at all, so maybe I shouldn’t have seen her at all until I turned it on.
A third AI-powered chatbot, Chad, is also in development , and I can imagine what he’ll be like based on the name alone.
At worst, it’s another step toward parasocial AI , with the same potential for catastrophic failure as regular AI, the same problems as regular AI , and some perhaps manipulative gamification elements. But I feel like the problems with the robot girl are well-worn, so I won’t go into them. Instead, even at best, I’m still left wondering what the point of all this is.
I love anime , but I don’t see a robotic maniac girl who can’t be crossed off as deserving more than a few minutes of my time. It’s basically just unattributed Google and an annoying filter I have to wade through to get to anything important. And even then, I had to wear headphones so my neighbors couldn’t hear it. Maybe I would have gotten more out of it at 14, but at 30, it’s an awkward way to spend my time that mostly leaves me confused. Targeting a niche audience of gooners isn’t how I’d rebuild the entire internet, but judging by Musk’s reposts , maybe I’m just not smart enough.
If you have a little more imagination than I do, this might be more than a little new to you. But since AI is still trying to establish itself as the next big thing, I’ll say that “cringe” isn’t the first thing I’d draw for it. These animated avatars are the closest I’ve seen to an NFT AI in a while, and given the way things are right now , that doesn’t bode well for the future of the industry.