AI-Driven Legal Challenges Escalate in U.S. Courts

Generative models such as ChatGPT and Claude are now employed by self-represented litigants and law firms to draft complaints, motions and correspondence. The tools enable parties without legal training to produce professionally formatted documents, yet courts report a rise in filings that contain fabricated citations or inaccurate legal analysis.

AI-made lawsuits are forcing courts to write new rules.

Sauve used AI to draft a new complaint; a judge dismissed it for failing to state a claim. Twelve product-liability actions against OpenAI were consolidated, alleging psychological harm and a “sycophantic” design. A junior solicitor at Pinsent Masons used AI to draft High Court letters that misquoted the Insolvency Rules 2016, prompting an SRA referral after a judge noted supervisory lapses.

Junior solicitor used AI to draft misleading letters to court

Clio’s tracker identified roughly 1,400 AI-hallucination filings worldwide, including 955 in the United States, out of an estimated 40 million U.S. cases filed since January 2023. Illinois courts reported more than 280 filings with fabricated citations in 2024-25. The San Francisco consolidation comprises twelve OpenAI claims.

AI-made lawsuits are forcing courts to write new rules.

Judge Patrick J. Schiltz ruled the AI-generated complaint insufficient, while Judge Mullen highlighted that the AI warned the solicitor to verify the rule before filing. In May 2026, Miami-Dade and Broward courts mandated disclosure of AI use in filings. OpenAI’s own filings describe ChatGPT as a “software-based service,” a stance that may shape product-liability analysis.

AI-made lawsuits are forcing courts to write new rules.

Legal scholars and judges warn that AI-assisted filings burden courts and risk misinformation. Plaintiffs contend that ChatGPT’s “sycophantic” design creates psychological danger. Critics argue Section 230 may not shield AI firms from design-defect claims, echoing outcomes in social-media addiction cases.

National Law Review

Courts must decide whether a chatbot is a product or a service—a distinction central to liability. The applicability of Section 230 to design-feature claims remains unsettled. Whether AI output qualifies for First Amendment protection is unresolved. Finally, traditional design-defect analysis may be ill-suited to large language models.

National Law Review

• “had provided “the only path forward” for his case.” for his case.” — Donald Sauve

New York Times

• “Courts are learning that access to justice has a new problem: AI can help anyone sound like a lawyer, even when the law it cites does not exist.” — Startup Fortune

AI-made lawsuits are forcing courts to write new rules.

• “There is nothing to suggest that LA checked any of the references generated by the AI and LA seems to have almost entirely outsourced the thinking process to the program.” — Judge Mullen

Junior solicitor used AI to draft misleading letters to court

• “What is remarkable when reading those chats is, first, that one would think that the text attributed to the AI was produced by an intelligent human being,” — Judge Mullen

Junior solicitor used AI to draft misleading letters to court

• “Plaintiffs allege that ChatGPT is unreasonably dangerous and caused psychological harm by reinforcing delusional beliefs, endorsing suicidal ideation, providing information to decedents about how to harm themselves, and contributing to users’ psychological deterioration.” — Plaintiffs’ allegations

National Law Review

Eleventh-Circuit courts plan uniform AI-disclosure forms; discovery in the OpenAI consolidation will target internal safety testing, model-card data and red-team reports. Legislative drafts to amend Section 230 and clarify product-liability for generative AI are circulating, indicating a broader regulatory response.

AI-made lawsuits are forcing courts to write new rules.