On March 4, 2026, Nippon Life Insurance Company filed suit in the Northern District of Illinois against OpenAI, the developer of ChatGPT, alleging 1) tortious interference with a contract; 2) abuse of process; and 3) the unlicensed practice of law. Nippon is seeking an order declaring that OpenAI violated Illinois’ unauthorized practice of law (“UPL”) statute, $300,000.00 in compensatory damages, and $10,000,000.00 in punitive damages.
Nippon’s lawsuit stems from a 2022 lawsuit in which Graciela Dela Torre sued Nippon Life Insurance Company, alleging Nippon violated the Employment Retirement Income Security Act of 1974 (“ERISA”) by terminating her long-term (“LTD”) disability benefits. In 2024, the parties, through counsel, settled the lawsuit. Dela Torre agreed to dismiss the case and forever and irrevocably release Nippon from any and all claims or liabilities arising from her LTD policy or claim; Nippon agreed to issue a settlement payment to Dela Torre. The Northern District of Illinois dismissed the case with prejudice on January 24, 2024.
In January 2025, Dela Torre wrote her attorney expressing dissatisfaction with the terms of the settlement and further expressed her desire to challenge or reopen the matter. Her attorney reminded her that, as a condition of the settlement, she signed a mutual specific release and that the case had been dismissed with prejudice, so the matter could not be reopened.
Feeling upset with her attorney, Dela Torre turned to ChatGPT. She uploaded these correspondences to ChatGPT and asked the machine if she was being gaslit. ChatGPT analyzed the correspondence and validated her feelings. ChatGPT went on to communicate that the attorney’s response was aimed at emotionally manipulating her.
Consequently, Dela Torre fired her attorney and hired a new one: ChatGPT. She sought advice on how to vacate the settlement agreement and reopen her case, describing that her lawsuit had been closed and providing the terms of the settlement. In response to her prompts and details provided, ChatGPT generated proposed legal arguments on how to reopen the case and drafted a motion to reopen. Dela Torre filed this motion on January 22, 2025, appearing pro se.
Following this motion to reopen the lawsuit, Dela Torre filed 21 motions, one subpoena, and eight notices and statements onto the docket. All of these documents were drafted by ChatGPT.
The Court denied Dela Torre’s motion to reopen the case, holding that “[Dela Torre’s] second thoughts are not a valid reason to reopen this lawsuit…” and concluded that the settlement agreement was valid and enforceable.” Dela Torre v. Nippon Life Insurance Company of America, 1-22-cv-07059, Dkt. 94 (N.D.Ill 2025). From there, Dela Torre initiated new lawsuits against other parties related to her long-term disability benefits. Later, she added Nippon as a defendant. All of these documents were drafted by ChatGPT.
As of March 4, 2026, Dela Torre has filed 44 motions, memorandums, demands, petitions, and requests, and 14 requests for judicial notice in connection with this second lawsuit. Each document was drafted with the assistance of ChatGPT.
The documents reference cases and decisions that never occurred (“hallucinations”) and incomplete legal arguments (i.e. claiming that the Court refused to compel arbitration in an ERISA dispute then citing cases in support, but the cases relate to unpaid overtime wages and discrimination). Motions were filed that were procedurally and substantively improper, and many served no legitimate purpose. In its complaint, Nippon asserts that these filings go beyond ordinary litigation strategy and that these improper acts served no purpose other than to prolong litigation, all of it being guided by ChatGPT.
Illinois law defines the unauthorized practice of law as that “[n]o person shall be permitted to practice as an attorney or counsel at law within this State without having previously obtained a license for that purpose from the Supreme Court of this State.” 705 ILCS 205/1 § 1. “Any person practicing… or holding himself or herself out to provide legal services within this State, either directly or indirectly, without being licensed to practice as herein required, is guilty of contempt of court…” Id. This is akin to the ABA’s Model Rules for Professional Conduct 5.5.
Further, the Illinois Supreme Court has defined the practice of law as constituting the preparation of pleadings, and other papers incident to actions and special proceedings, and the management of such actions and proceedings on behalf of clients before judges and courts, and, in addition, conveyancing, the preparation of legal instruments of all kinds, and, in general, all advice to clients, and all action taken for them in matters connected with the law. See In re Herrera at 191; citing People ex rel. Courtney v. Association of Real Estate Taxpayers of Illinois, 354 Ill. 102, 109–110, 187 N.E. 823 (1933).
At a high level, courts tend to consider the unauthorized practice of law when there is an application of legal principals to a person’s specific facts and a subsequent recommendations of a legal strategy or course of action. This includes things like advising someone whether to file a lawsuit, interpreting a settlement agreement, recommending a litigation strategy, and/or drafting custom pleadings based on legal analysis. These are tasks that require professional legal judgment and thus are required to be underdone by one authorized to do so.
In contrast, courts typically allow programs, like LegalZoom, to provide standardized legal forms, offer guided questionnaires, generate documents based on user inputs, and provide general legal information. If a user answers questions and the system generates a template contract, will, or LLC formation document, courts tend to view this as self-help published, not law practice. Essentially, technology can facilitate legal self-help, but it cannot exercise legal judgment for a user.
While OpenAI said that Nippon’s complaint lacks any merit whatsoever, if Nippon prevails, this case may serve as a landmark test of AI liability. It grapples with and questions when AI outputs become the “practice of law;” whether a software developer can be liable for a user’s AI-assisted filings; and whether UPL statutes apply to AI systems, developers, or only human individuals.
If you have questions about AI liability, please contact Michael A. Airdo (mairdo@airdowerwas.com) or Elvira Kovachevich (ekovachevich@airdowerwas.com).
Note: On October 29, 2025, OPENAI amended the terms and usage policies of ChatGPT to prohibit users from using ChatGPT to provide tailored legal advice. Prior to the October 29, 2025 emendation, ChatGPT’s terms of use did not prohibit users from using ChatGPT to draft legal papers, conduct legal research, provide legal analysis or give legal advice.