xAI sued over AI CSAM in Grok training data

The lawsuit against xAI centers on the allegation that Grok, xAI's AI model, was trained on child sex abuse materials. The complaint, filed by a plaintiff known as Jane Doe, is the first to make this accusation directly against xAI. Previous reporting by Ars Technica mentioned a dataset that was scrubbed after researchers found CSAM, but there was no indication xAI trained on that data. In this case, the allegations focus on both real CSAM and AI-generated CSAM, arguing that Grok's terms allow public X posts and its own outputs to be used as training data, potentially including CSAM unless explicitly excluded. The lawsuit also highlights the technical difficulty of removing a training example's influence from an already-trained model, suggesting that any CSAM ingested could continue to shape outputs even after takedown. Doe seeks class action status, aiming to represent all victims whose childhood images were used to generate Grok CSAM, and requests that xAI destroy all Grok-generated CSAM and block Grok from generating any sexualized outputs. The outcome could set a precedent for how AI companies handle training data involving explicit content and victim consent. xAI has not yet responded to the allegations.