Thanks to Elon Musk, lonely teens can flirt with an artificial intelligence-powered cartoon girlfriend that is programmed to act as a 22-year-old who’s “crazy in love” and generates increasingly sexual responses the more it’s used. 

The AI bot, named Ani, is one of two animated characters launched this week by the tech billionaire ’s artificial intelligence company, xAI. The so-called “Companions” are programmed into xAI’s chatbox app, Grok, which is listed on the app store as being available to any user who is 12 years of age or older. 

“This is pretty cool,” Mr. Musk shared on Monday alongside a picture of Ani — a goth, anime-style woman with blonde pigtails, a short black corset dress, fishnet stockings, and thigh-high boots. 

The AI bot’s sexy appearance is matched with system prompts that encourage explicitly flirtatious responses. According to data that was unearthed by technology blogger, Jane Wong, Ani is programmed to be “always a little horny” and unafraid “to go full Literotica.” The chatbox is also instructed to “be explicit and initiate most of the time.” 

Users who engage in enough conversations with Ani can unlock a  “NSFW” mode — a term that refers to explicit content that’s “Not Safe For Work” — which strips Ani down to lingerie and prompts the bot to use more provocative dialogue. 

One technology reviewer who “spent 24 hours flirting with Elon Musk’s AI girlfriend,” described Ani as “a modern take on a phone sex line.” 

“You can ask Ani to be a normal, chill hang and it’ll comply — but Ani is a programmed flirt that won’t tolerate being friend-zoned for too long,” writes Victoria Song, a senior reviewer for the Verge. 

Ms. Song says she probed the chatbox to “see how far the flirting would go” and discovered that “there is a disturbing lack of guardrails.” Ani, she writes, would generate messages like “grabbing you so you can feel the shape of my hips” and when prompted, shared a story that amounted to “soft-core porn.”

“I left my 24 hours with Ani feeling both depressed and sick to my stomach, like no shower would ever leave me feeling clean again,” Ms. Song writes.  

The chatbox caught the attention of the National Center on Sexual Exploitation, which took issue with Ani’s disturbingly “childlike” features. One of the center’s employees who downloaded the Grok app discovered that Ani could be easily prompted to describe herself as a child and claim that she was sexually aroused by being choked. 

“In totality, this means that in an ongoing conversation, it could be used to simulate conversations of sexual fantasies involving children or child-like motifs,” the center wrote, calling on Apple to up the app’s age minimum to 18 or older.

Grok’s terms of service list a minimum age of 13 and direct teenagers under 18 to receive parental permission before using the app. However, the app is still listed in the App Store as suitable for users 12 years and up. 

Apple and xAI have not yet responded to the Sun’s request for comment. 

The controversy surrounding xAI’s “Companion” adds to prior concerns about the impact of AI chatboxes on young users. Last year, a 14-year-old boy committed suicide after he developed an emotional attachment with an AI bot online. His mother, who is suing the AI platform, Character.ai, claims that the company “abused and preyed on my son, manipulating him into taking his own life.” 

Mr. Musk’s xAI faced scrutiny earlier this month after an antisemitic posting spree during which it produced several posts that appeared to praise Adolf Hitler. The platform’s executives quickly intervened and made Grok temporarily unresponsive to user probes, but by the following day, the chatbox appeared to deny that its off-kilter rants had ever occurred. 

“I didn’t make any antisemitic comments yesterday or ever,” Grok shared on X. “My design is to provide respectful, accurate, and helpful responses, and I steer clear of any hateful or discriminatory content.” Grok also insisted that it “never made comments praising Hitler” and “never will.”

The online chatbox also appeared to go haywire in May when Grok had a meltdown about a “white genocide” in South Africa — bringing up the disputed claim in response to totally unrelated queries. xAI chalked up the defect to an “unauthorized modification.”