Elon Musk said Wednesday that he was “not aware of any images of naked minors generated by Grok,” hours before California’s attorney general launched an investigation into xAI’s chatbot for “spreading non-consensual sexually explicit content.”
Musk’s denial comes amid mounting pressure from governments around the world, from the UK and Europe to Malaysia and Indonesia, after X users began asking Grok to transform photos of real women and even children into sexualized images without their consent. Copyleaks, an AI detection and content governance platform, estimated that approximately one image is posted every minute on X. Another sample collected from January 5th to January 6th found that 6,700 images were posted per hour over a 24-hour period. (X and xAI are part of the same company.)
“This material…is being used to harass people on the internet,” California Attorney General Rob Bonta said in a statement. “I urge xAI to take immediate action to prevent this from happening again.”
The AG’s office will investigate whether and how xAI violated the law.
Several laws exist to protect the subjects of non-consensual sexual images and child sexual abuse material (CSAM). Last year, the Take It Down Act was signed into federal law, making it a crime to knowingly distribute non-consensual intimate images (including deepfakes) and requiring platforms like X to remove such content within 48 hours. California also has its own set of laws that Gov. Gavin Newsom signed in 2024 to crack down on sexually explicit deepfakes.
Toward the end of the year, Grok began responding to requests from users on X to create sexually explicit photos of women and children. The trend appears to have started after certain adult content creators encouraged Grok to generate sexual images of themselves as a form of marketing, and other users subsequently issued similar prompts. In a number of public lawsuits involving celebrities like “Stranger Things” actress Millie Bobby Brown, Grok responded to prompts to alter photos of real women, including altering their clothing, body position, and physical characteristics in overtly sexual ways.
According to some reports, xAI has started implementing safeguards to address this issue. Grok requires a premium subscription before responding to certain image generation requests, and may still not generate images. April Cozen, Copyleaks’ vice president of marketing, told TechCrunch that Grok may respond to the request in a more general or subtle way. They added that Grok appears to be more tolerant of adult content creators.
tech crunch event
san francisco
|
October 13-15, 2026
“Overall, these behaviors suggest that, although discrepancies remain, X is experimenting with multiple mechanisms to reduce or control the production of problematic images,” Cozen said.
Neither xAI nor Musk has publicly addressed this issue head-on. A few days after the instance began, Musk appeared to downplay the issue by asking Grok to generate an image of himself in a bikini. On January 3, X’s safety account announced that the company would take “actions against illegal content on X, including (CSAM),” but did not specifically mention Grok’s apparent lack of safety measures or the creation of sexualized and manipulated images involving women.
This positioning mirrors what Musk posted today, highlighting illegality and user behavior.
“I had no knowledge of any images of naked minors produced by Grok. Literally zero,” Musk wrote. This statement does not deny the existence of bikini photos or sexualized edits more broadly.
Michael Goodyear, an associate professor at New York Law School and a former litigator, told TechCrunch that Musk likely focused on CSAM because of the heavy penalties for compiling or distributing sexual images of children.
“For example, in the United States, a distributor of CSAM or someone who threatens a distributor can be sentenced to up to three years in prison under the Take It Down Act, compared to two years in prison for non-consensual adult sexual images,” Goodyear said.
He added that the “bigger point” was Musk’s attempt to draw attention to problematic user content.
“Clearly, Grok does not spontaneously generate images; it only generates them upon user request,” Musk wrote in the post. “When asked to generate an image, Grok refuses to generate anything illegal. Grok’s operating principle is to follow the laws of a particular country or state. Hostile hacks against Grok prompts can sometimes lead to unexpected things. In such cases, we fix the bug immediately.”
In summary, this post characterizes these incidents as unusual, attributed to user requests or adversarial prompts, and presents them as technical issues that can be resolved with fixes. We have yet to realize Grok’s fundamental safety design flaw.
“Regulators may be mindful of free speech protections and consider requiring AI developers to take proactive steps to prevent such content,” Goodyear said.
TechCrunch reached out to xAI to ask how many instances of non-consensual sexually manipulated images of women and children it has caught, what specific guardrails it has changed, and whether the company has notified regulators of the issue. TechCrunch will update this article if companies respond.
The California AG is not the only regulator trying to hold xAI accountable for this issue. Both Indonesia and Malaysia temporarily blocked access to Grok. India required X to immediately make technical and procedural changes to Grok. The European Commission has ordered xAI to preserve all documents related to the Grok chatbot as a prelude to opening a new investigation. And Ofcom, the UK’s online safety watchdog, has launched a formal investigation under the UK’s Online Safety Act.
xAI has previously come under fire for sexually explicit images of Grok. As AG Bonta pointed out in a statement, Grok includes a “spicy mode” for producing explicit content. An October update made it even easier to jailbreak with few safety guidelines, leading many users to use Grok to create hardcore pornography and graphic, violent sexual images.
Many of the pornographic images Grok creates are of people generated by AI, and while they may still seem ethically questionable to many, the harm to the individuals depicted in the images and videos is probably less severe.
“When we allow AI systems to manipulate images of real people without their explicit consent, the consequences can be immediate and deeply personal,” Copyleaks co-founder and CEO Alon Yamin said in an emailed statement to TechCrunch. “From Sora to Grok, we are seeing rapid improvements in AI capabilities for manipulated media, which requires detection and governance to prevent abuse now more than ever.”
