On March 4, Nippon Life Insurance brought what may be a novel legal question to the U.S. District Court for the Northern District of Illinois. Nippon is suing OpenAI, which designs and runs ChatGPT, alleging that ChatGPT is practicing law without a license. The company claims that OpenAI’s product has cost it hundreds of thousands of dollars in legal work by supplying a pro se litigant with dozens of frivolous filings.
Some legal critics say that Nippon’s particular complaint may not succeed. However, the use and misuse of ChatGPT and other LLMs (large language models) is an urgent concern, not just for courts but for doctors, architects, artists, and almost every professional with specialized knowledge. Can attorneys help bring about safeguards for the use of artificial intelligence?
The Company’s Allegations: A Vexatious Litigant with a Chatbot
According to the statement of facts in Nippon’s March 4 complaint, the trouble stems from a 2021 disability insurance dispute with an employee. Graciela Dela Torre, who suffered from carpal tunnel syndrome and other ailments, alleged Nippon had wrongfully terminated her long-term disability benefits. She agreed to a settlement with Nippon in 2024, waiving further claims against the company.
A year later, Dela Torre contacted her attorney, Kevin Probst, and expressed dissatisfaction with the settlement. Attorney Probst disagreed with her reasoning and reminded her that the settlement agreement was final. Dela Torre then took her attorney’s response to OpenAI’s ChatGPT.
According to the complaint, she asked the chatbot “whether she was being gaslighted. ChatGPT analyzed the response and determined that Mr. Probst’s response invalidated Dela Torre’s feelings … ChatGPT ultimately concluded that the tactics used in [his] response constituted gaslighting and were aimed at emotionally manipulating Dela Torre.”
Dela Torre fired her attorneys and began to act pro se. She filed a motion attempting to reopen her lawsuit, and when that failed, she expanded on her arguments in another lawsuit naming yet more parties. Dozens of motions, requests, and filings followed, some based on events that had not actually happened. All of them appear more or less directly taken from ChatGPT’s outputs.
In this complaint, Nippon has not named Dela Torre as a defendant. Instead, it is suing OpenAI alone for economic and punitive damages on three counts: abuse of process, tortious interference with contract, and unlicensed practice of law.
The Law and AI Use: Litigation, Legislation, or Both?
Nippon argues that “as Dela Torre’s legal assistant and advisor, OpenAI intentionally induced and facilitated Dela Torre’s breach of a valid and enforceable settlement agreement” and “aided and abetted her abuse of the judicial process.” However, legal commentators have suggested that Nippon is unlikely to succeed in its suit against OpenAI, at least with this complaint.
Could one really impute such specific knowledge and intention to OpenAI? To ChatGPT itself? Nippon’s complaint also makes it clear that Dela Torre had a personal grievance before her AI use and throughout her later litigation. She, not ChatGPT, filed the motions, but she is not named in their complaint.
Eran Kahana, an AI attorney and Research Fellow at Stanford, opines that what Nippon actually has is a product liability claim. OpenAI has recently modified its rules to forbid the use of ChatGPT for legal advice, but disclaimers cannot always protect a company from a product liability lawsuit if they failed to warn consumers adequately. And according to Kahana, “a manufacturer cannot disclaim its way out of a design defect that makes the product unreasonably dangerous for its foreseeable use.”
Should this be a legislative matter? State lawmakers are already working to address liability issues with AI tools. Legislators in New York are considering a bill that would expressly make companies liable for the unauthorized practice of law by their AI chatbots. However, the current administration favors a “national policy framework” for AI use, and it may challenge state-level AI legislation that is not “minimally burdensome” to businesses.
Nonetheless, creating legal guardrails for AI use and abuse has had bipartisan support in Congress. As more litigation appears about artificial intelligence use for sensitive matters, particularly mental health issues, laws governing AI use will develop in one way or another.
Related Resources
- SDNY Judge Rules Defendant’s Own Use of AI Tools for Strategizing Not Privileged (FindLaw's Practice of Law)
- Judge Strikes Hallucinated Citations From Brief With No Leave to Amend (FindLaw's Practice of Law)
- New Software for SCOTUS to Detect Financial Conflicts for Justices (FindLaw's Federal Courts)
