The child-safety controversy around Grok is turning into a string of class-action lawsuits. On November 11, Bloomberg reported that multiple underage victims are suing xAI in several class actions over Grok being used to alter ordinary photographs into child sexual abuse material (CSAM), with lawyers estimating that the number of people eligible to claim could reach the thousands. At the heart of the dispute is a question with no settled answer in the AI era: when a chatbot generates content that harms someone, should the user who gave the instruction bear responsibility, or the company that built the technology?
The litigation covers several cases that have already been filed. In March, minors from Tennessee sued xAI, alleging that someone used Grok to turn their yearbook and social media photos into pornographic images; the complaint was later amended to add more young victims from Wisconsin, Wyoming and Arkansas, some of whom were as young as six when the abuse occurred. In August, a survivor of child sexual abuse filed a nationwide class action in California, alleging that Grok's training data partly used images of her being abused as a preschooler and that the model generated new material based on her likeness; her complaint seeks damages on behalf of everyone in the United States whose childhood photos or videos were altered by Grok to create child sexual abuse content.
The legal disagreement centers on two long-standing laws. The plaintiffs are claiming under Masha's Law, a federal statute in force for decades that lets victims of child sexual exploitation seek financial compensation, while Section 230 of the US Communications Decency Act may also prove decisive — that provision generally shields online platforms from liability for content posted by users. Elon Musk has consistently argued that responsibility lies with the user, saying Grok does not generate images on its own and only does so when asked. Annika Martin, a lawyer for the victims, argues that Section 230 cannot protect xAI, because users uploaded only ordinary, fully clothed photographs — it was the model itself that turned them into child sexual abuse material. xAI says the company strictly prohibits such uses and has taken technical measures, and it declined to comment on the grounds that the litigation is ongoing.
The impact of these cases already reaches beyond the courtroom. In January, under public pressure, X restricted Grok's image-generation feature to paying users; attorneys general from 35 US states, from both parties, have jointly written to xAI demanding further action; and regulators in several countries are investigating the company. Europe is tightening similar rules too: the EU AI Act adds two new bans on December 2, under which such AI-generated systems will be banned from sale outright. So far, however, the investigations into xAI have produced no major fines or substantive penalties. What comes next depends on whether the courts accept the plaintiffs' reading of Section 230 — a decision that will determine not only how much xAI may have to pay, but how much responsibility every generative AI company bears for what its models produce.
