ChicagoAI citations sank a US doxing suit.
D’Ambrosio sued to unmask women who reviewed him negatively; the court found AI-fabricated citations throughout the brief.
MarcTrent.AI has until June 16 to respond on sanctions; the case was dismissed with prejudice.
Sources: Ars Technica
THE HALLUCINATION PROBLEM
Large language models predict plausible-sounding text, not true text. When asked for case citations, they generate strings that look exactly like real citations — correct reporter format, plausible court, plausible judge — but reference cases that do not exist. The model is not lying; it has no concept of truth to lie about.
RULE 11
Federal Rule of Civil Procedure 11 requires that every filing be backed by a reasonable inquiry into facts and law. Signing a brief is a personal certification by the attorney. Citing cases that do not exist is the textbook violation — judges have imposed sanctions for it since the rule was tightened in 1983.
THE MATA PRECEDENT
In June 2023, a Manhattan federal judge sanctioned two lawyers in Mata v. Avianca $5,000 each for submitting a ChatGPT-generated brief with six fabricated cases. It became the canonical warning shot — and a steady stream of copycat sanctions has followed in courts from Texas to Colorado to the UK.
WHY SECTION 230 IS THE BACKDROP
Doxing suits against review platforms typically die at the threshold: Section 230 of the Communications Decency Act immunizes interactive computer services from liability for user-posted content. To get past it, a plaintiff must usually unmask the user and sue them directly — which is what the subpoena fight in this kind of case is really about.
DISMISSED WITH PREJUDICE
A dismissal with prejudice is final — the plaintiff cannot refile the same claim. Courts reserve it for cases where amendment would be futile or where the plaintiff's conduct (frivolous filings, fabricated evidence, bad-faith litigation) forfeits the chance. Pairing it with a sanctions order signals the court viewed the misconduct as severe.