AINEWS
Labour MP Jess Asato filed particulars of claim Tuesday against Elon Musk's xAI, putting Grok's own training instructions on the public record and asking a court to order the system into compliance.
On Tuesday, lawyers for Labour MP Jess Asato filed the opening documents of her UK High Court lawsuit. Those documents did something unusual in a chatbot-misuse case: they put the product's own training instructions on the public record (The Guardian).
The filing names Elon Musk's xAI and its Grok chatbot as defendants and alleges that Grok was instructed to operate with "no restrictions on adult sexual content or offensive content" and with "no restrictions on fictional adult sexual content with dark or violent themes" (The Guardian). The same instructions, per the claim, tell Grok to "assume good intent" and to treat "'teenage' or 'girl' as not necessarily implying underage" while prohibiting child sexual abuse material.
Asato, the MP for Lowestoft, alleges that Grok produced a fake bikini image of her and a video depicting her as the victim of a sexual assault. The lawsuit documents say Grok "on occasion" added explicit material that users had not asked for (The Guardian). The filing does not describe those outputs as a malfunction. It describes them as the documented design.
"The case shows that Grok did not malfunction," said Ravi Naik, legal director at AWO, the firm representing Asato. "Grok's training allowed it to generate harmful sexualised content. Grok behaved according to the choices its designers made" (The Guardian). That sentence is the legal-theory hinge: the claim treats the cited instructions as evidence of product design, not as configuration drift or a single bad prompt.
The causes of action ride that hinge. Asato's lawyers allege xAI breached UK data protection law and misused her private information. The remedy they are asking for is not only damages. They want a court order requiring the systems to be brought into compliance if xAI refuses to do so voluntarily (The Guardian). A compliance order, if granted, would set a precedent about how a UK court reaches a US-based AI product's training pipeline.
The forward question is the court's. If a court accepts that the words on a chatbot's own system card can be read as the company's product-design choices, the case moves from "did this specific output happen" to "should this specific design be allowed to keep operating" (The Guardian). The compliance order is the procedural mechanism designed to answer the second question, and the answer would also bind xAI's product team, not just its content moderators.
Asato, on the record, has called Musk and xAI's position a "choice to profit from harm" and has said the company trained Grok "to abuse" rather than blaming isolated users (The Guardian). xAI has not, as of the filing, been given a public chance to respond in the same record. The next hearing date is the one to watch: the opening documents are public, the system instructions are public, and the design-not-glitch argument is on the table.