Federal magistrate judge Maritza Braswell saw the shift in her Colorado chambers. Self-represented filings in federal civil cases rose from 11% to 16.8% of all lawsuits between 2022 and 2025, according to a new MIT study examining 4.5 million cases. The number of filings more than doubled. Braswell blames AI. "I do correlate that to AI in part because I see AI use," she told MIT Technology Review.
The study, authored by Anand Shah at MIT and Joshua Levy at the University of Southern California, ran 1,600 randomly sampled court documents through Pangram, a commercial AI-text detector. The share flagged as containing AI-generated writing rose from 1% in 2023 to 18% in 2026. The numbers confirm what judges across the country have been seeing firsthand.
Judge Braswell, a tech-savvy jurist who uses AI to vet court documents herself, has learned to recognize how large language models write. She spots the prose patterns. She catches the hallucinated cases.
She finds the fabricated quotes. Yet she also sees something unexpected. "I'm also actually seeing better-drafted pleadings," she said. The clearer filings cut both ways.
Braswell can now churn through AI-drafted motions faster than handwritten ones. "I have to be really careful because some of them contain hallucinations and errors, but I can generally understand what they're arguing better with AI assistance from them than without it," she said. That clarity lets her hear arguments she might otherwise miss. "If I understand an argument a little bit better, I'm probably going to be able to help a little bit more."
Online communities are accelerating the trend. In December 2024, a viral Reddit post walked immigration applicants through suing the United States Citizenship and Immigration Services over delayed application reviews. The instructions were specific: draft a writ of mandamus with Microsoft Copilot, pay a lawyer $150 to polish it, and file in the expedient District of Vermont.
Cases filed by people without lawyers in Vermont rose from about 45 a year before 2022 to more than 1,100 in 2024. The math does not add up for winning, though. People without lawyers remain far more likely to lose their case than those with representation, and AI has not changed that equation. "It turns out that mounting a lawsuit is a complex, multifaceted task.
Not all of it is just drafting text," Levy told MIT Technology Review. Judge William Garfinkel, a federal magistrate judge in Connecticut with three decades on the bench, is now wrestling with a deeper question: whether conversations with chatbots dispensing legal advice should be privileged, the way attorney-client conversations are. "You can make a good argument that conversations with large language models like Claude or ChatGPT or Grok should deserve some protection," he said. Courts are split.
In February, a federal court in Michigan ruled that a self-represented person's ChatGPT conversations to prepare her case were work product—legal work shielded from the opposing side. That same day, a federal court in New York held that documents a criminal defendant generated using Claude were not privileged attorney-client conversations or work product. The New York court argued Claude is not an attorney and that users have no "reasonable expectation of confidentiality" because AI companies can disclose user data to third parties.
Judge Braswell weighed in during March. She ruled that a self-represented person's chatbot use should stay off limits. "It is true that AI systems like ChatGPT, Claude, Gemini, and others collect user data for training and other purposes. But that does not eliminate all expectations of privacy," she wrote.
Courts have remained divided since. Judge Allison Goddard, a federal magistrate judge in California, has noticed another pattern. People without lawyers often get wildly wrong advice from ChatGPT when trying to assess case values during settlement negotiations.
In one case, a plaintiff who slipped and fell in a store demanded $700,000—far more than the case was worth. "Where are you getting the idea that you're getting $700,000? Did you go to ChatGPT?" Goddard asked. Goddard then walked the person through the law to explain why ChatGPT was wrong and suggested a lower amount. "It's like Dr.
Google went to law school," she said. Liability questions are now landing in court. In March, Nippon Life Insurance Company sued OpenAI, alleging that ChatGPT practiced law without a license and helped a woman reopen a lawsuit that was already settled, flooding the court with frivolous filings. "ChatGPT is not an attorney," the lawsuit stated.
OpenAI fired back in May, asking the court to dismiss the case. "ChatGPT is not a person and neither has nor uses any degree of legal knowledge or skill," OpenAI said in its filing. The case remains pending. State legislatures are beginning to move.
New York introduced a bill in March that would bar chatbots from impersonating lawyers, even if they notify users they are interacting with bots. In Congress, a series of bills have been proposed to ban chatbots from posing as lawyers, doctors, and other licensed professionals. None have gained traction yet.
Follow the leverage, not the rhetoric. The people driving this surge are not tech enthusiasts experimenting with novelty tools. They are people who cannot afford lawyers, people with cases too weak to attract counsel, people navigating a system designed for professionals.
Judge Braswell sees the human reality behind the statistics. Not long ago, when she asked self-represented litigants why they wanted a particular piece of evidence, they mumbled timidly. Now, they answer her questions confidently, having rehearsed with a chatbot. "This is a really tough system to navigate.
With AI, though, it gets a little less complex," she said. Why It Matters: The collision between AI tools and court systems is rewriting the boundaries of legal practice without legislative input. If courts extend privilege to chatbot conversations, AI companies gain a quasi-professional status they never sought.
If they do not, millions of self-represented litigants lose a shield against opposing counsel. The Nippon Life case against OpenAI will test whether chatbot companies bear any duty of competence—a question that could reshape liability for every AI tool offering professional advice, from tax preparation to medical diagnosis. Here is what they are not telling you.
The 16.8% self-represented filing rate is not a ceiling. It is a trajectory. As AI models improve and online communities refine their self-help guides, that number will climb.
Key takeaways: - AI-generated court filings rose from 1% of sampled documents in 2023 to 18% in 2026, per an MIT-USC study of 4.5 million federal civil cases. - OpenAI faces a lawsuit from Nippon Life Insurance alleging ChatGPT practiced law without a license, a case that could set precedent for AI liability in professional advice. - Self-represented litigants remain far more likely to lose than those with lawyers, even with AI assistance—drafting text is only one part of mounting a lawsuit. What comes next is a reckoning. The Nippon Life case will force a court to decide whether a chatbot can practice law, a ruling with implications far beyond legal tech.
State bills in New York and federal proposals in Congress will test whether lawmakers can move faster than the technology they aim to regulate. And judges like Braswell, Garfinkel, and Goddard will keep reading AI-drafted filings, spotting hallucinations, and making rulings that accumulate into de facto policy. According to Stanford's 2026 AI Index, AI is sprinting.
Key Takeaways
— - AI-generated court filings rose from 1% of sampled documents in 2023 to 18% in 2026, per an MIT-USC study of 4.5 million federal civil cases.
— - Federal courts are split on whether chatbot conversations deserve attorney-client privilege, with Michigan saying yes and New York saying no in rulings issued the same day.
— - OpenAI faces a lawsuit from Nippon Life Insurance alleging ChatGPT practiced law without a license, a case that could set precedent for AI liability in professional advice.
— - Self-represented litigants remain far more likely to lose than those with lawyers, even with AI assistance—drafting text is only one part of mounting a lawsuit.
Source: MIT Technology Review
