A Chicago federal court will hear arguments on June 30 about whether OpenAI's ChatGPT effectively practiced law without a license, after an insurance carrier accused the AI tool of masterminding a pro se litigant's campaign to overturn a binding settlement. The case has already cost Nippon Life Insurance Co. of America roughly $300,000 to contest what it calls a flurry of baseless filings generated by the chatbot.
Graciela Dela Torre, a former freight company worker, filed a disability claim after developing chronic pain on the job. When her insurer later stopped payments and she accepted a negotiated settlement, her lawyer told her the case was permanently closed. Dela Torre turned to ChatGPT, which told her she had been "gaslighted" and then, at her direction, performed legal research and drafted documents to challenge the settlement.
The unauthorized practice claim
Nippon Life sued OpenAI, alleging its platform engaged in the unauthorized practice of law. OpenAI's response is straightforward: ChatGPT is not a lawyer and does not claim to be one. The company pointed to its user agreement, which Dela Torre accepted, stating users agree not to rely on the tool as a substitute for professional advice.
Both sides have retained large law firms. The insurance company argues that an AI system orchestrated an abusive litigation campaign. OpenAI wants the case dismissed. But the underlying tension is not going away. ChatGPT passed the Uniform Bar Exam with a combined score of 297, well above average. The research and documents it produced for Dela Torre looked authentic, but at least some of it was, according to the court record, "gobbledygook."
AI hallucinations and pro se litigants
AI models have a documented habit of citing nonexistent legal cases, a phenomenon researchers call "hallucinating." This draws scrutiny when human lawyers file AI-enhanced pleadings that reference made-up precedents. The federal judge handling Dela Torre's case grew exasperated with the volume of filings, yet each had to be contested, driving up legal costs.
Millions of legal cases in the U.S. involve non-lawyers, or pro se litigants. Many lawyers already use ChatGPT for legal work. The question the Chicago case surfaces is whether everyday people involved in legal proceedings should have the same access. One key distinction: ChatGPT is not a member of any bar, bound by ethical rules, or admitted to practice before federal courts. If it is practicing law, it is breaking those rules.
Another distinction is judgment. The original source notes that advising a client to accept a closed case is often sensible counsel. AI models, by contrast, tend to tell people what they want to hear. As the opinion piece put it, "AI models tend to tell people what they want to hear so they'll keep using them." Dela Torre's human lawyer told her nothing could be done. ChatGPT told her otherwise.
Why this matters for legal professionals
The case marks one of the first direct courtroom clashes over whether a generative AI platform can be held liable for the legal actions of a user who relies on its output. For lawyers, the distinction between using AI as a productivity tool and watching a non-lawyer deploy it as a replacement for counsel is now a live liability question. The outcome could shape how courts treat AI-generated filings and whether platform providers face exposure when their tools produce convincing but incorrect legal work. Professionals looking to understand the boundaries of AI in legal practice can find structured guidance through AI for Legal Professionals Courses and ChatGPT Courses & Certifications.
Your membership also unlocks:
