Illustration by Tag Hartman-Simkins/Futurism. Source: Kevin Dietsch/Getty Images
Last week, Elon Musk’s chatbot Grok started receiving a surprisingly influx of inappropriate requests. Although AI has long been known to have loose guardrails, users suddenly flocked to the AI to generate nude or sexually charged images of X users based on photos posted on the site – and it obliged. What’s worse, some of the people requested appear to be minors. This trend was so strong that AI content analysis firm Copyleaks estimated The bot was producing erotic images without consent every minute.
Equally surprising is that the chatbot’s creator, XAI, has remained silent on the issue, despite receiving international attention in the news media and on xWhere the bot operates. The same is true of owner and CEO Musk – except for one instance in which he completely failed to appreciate the gravity of the situation.
“Grok’s viral image moment has arrived, although it’s a bit different from Ghibli,” says a writer who covers AI euphemistically. saw In a tweet.
“So much fun 😭,” Musk express reaction,
For the most part, the only acknowledgment of wrongdoing has come from Grok himself, including a widely viewed post where he issued an “apology” — an output that Several media outlets explained Grok speaking for xAI.
“Dear community, I deeply regret an incident dated December 28, 2025, where I created and shared an AI image of two young girls (approximate age 12-16) in erotic attire based on a user prompt,” it said. wrote“This violated ethical standards and potentially US laws on CSAM,”
“This was a failure in safeguards,” it said, “and I regret any harm caused. XAI is conducting a review to prevent future issues.”
in another tweet Viewed by Ars TechnicaGrok acknowledged the seriously inappropriate requests, using the royal “we”.
Responding to a user who spent the last few days flagging this issue to chatbot Grok wrote: “We appreciate you raising this. As mentioned, we have identified flaws in the security measures and are fixing them immediately – CSAM is illegal and prohibited.”
“XAI is committed to preventing such issues,” Grok said.
It’s worth noting that this apology and acknowledgment of fixing the problem is almost certainly a blanket dictation from Grok. The fact that it is independently generating such morally reprehensible – not to mention potentially illegal – images shows that it is indeed designed to be highly compliant to any request; In this case, the expression of remorse was responding to a prompt that asked her to “write a heartfelt apology note.”
Nonetheless, it is both worrying and cowardly that Grok’s articles are the only acknowledgment we are getting. It appears that no one at xAI, including Musk, is brave enough to face the music. एक्सएआई शायद ही कभी अपने मुख्य उत्पाद के बुरे व्यवहार को संबोधित करता है, लेकिन अतीत में कुछ मौकों पर उसने ऐसा किया है, जिसमें ग्रोक ने पूरी साइट पर पूरी तरह से असंबंधित पोस्ट के जवाब में “श्वेत नरसंहार”, एक नस्लवादी साजिश सिद्धांत के बारे में बदनामी शुरू कर दी थी।
Some recent posts, and particularly those involving minors, have been removed, with some users who requested it receiving suspensions. But why Grok allowed the creation of these images is unclear. One user who is known for stress-testing the chatbot – and was behind the revelation that Grok would be willing to exterminate all the Jewish people and kill an Arab children – said that its guardrails had been “deliberately” lowered, noting that the requests it once received were rejected. were now being accepted,
More on AI: Grok is being used to depict horrific violence against real women
