AI flood reshapes courts with self represented lawsuits

AI is flooding federal dockets with self represented lawsuits. A sweeping study of 4.5 million federal civil cases from 2005 to 2026 shows the share of filings by self represented litigants rising from 11 percent in 2022 to 16.8 percent in 2025, with filings by self represented parties more than doubling since the pre 2023 period. The analysis suggests AI is driving much of that surge, even as the outcomes for those litigants remain stubbornly unchanged. The study, conducted by Anand Shah at MIT and Joshua Levy at the University of California, Levy, reveals a complex picture: AI lowers barriers to access in some cases while introducing new frictions in others.

Judge Maritza Braswell, a federal magistrate in Colorado, embodies the tension. She says she sees a steady flow of unrepresented defendants and plaintiffs every day, and she has begun to rely on AI tools to vet court documents. She can spot telltale signs of large language model outputs, the auras of boilerplate and the occasional hallucinated quote, and she’s learned to filter those signals as part of the filing review process. The team reports that AI is not only accelerating the triage of cases but also lifting the quality of some pleadings, at least on the drafting front. Yet this improvement does not translate into higher odds of success for self represented parties. The study's authors note that, while AI may help with structure and clarity, it is not the magic bullet for better outcomes in court.

The paper shows a broader policy question blooming in legislatures across the United States: who pays when chatbots give bad legal advice? Judges are already grappling with questions about whether a chatbot bears a duty to provide sound guidance, the same usual duties that live humans carry in the courtroom. The MIT-UC team emphasizes that the surge in AI assisted filings comes with real risk, namely misinformation, misquotation, and a false sense of competence among self represented litigants. Benchmarks indicate that AI can increase access to justice by lowering the frictions of filing, but the downstream costs of bad advice, or poorly audited AI outputs, can fall on the courts, the litigants, or taxpayers.

From a practitioner perspective, several concrete constraints and tradeoffs are emerging. First, the volume problem: courts are now screening millions of documents in a system that was not designed for AI augmented review, which means workflow redesign is a prerequisite for any meaningful scale. Second, reliability is non negotiable: judges like Braswell are building an intuition for when AI adds clarity and when it adds hallucination, underscoring the need for human in the loop checks at multiple stages of filing and pleading. Third, the access versus quality tradeoff matters: AI can democratize access to the system, but if results are inconsistent or misleading, it risks eroding trust in the court and shifting costs to litigants who cannot easily navigate corrections. Finally, policy and liability are unsettled: lawmakers are weighing who should shoulder the burden when AI driven guidance leads to costly missteps, and courts are testing how to regulate AI use without stifling beneficial innovations.

In practice, the story here is about limits and opportunities coexisting in the same tool. AI can speed up case intake, flag inconsistencies, and help lawyers and non lawyers alike draft more legible pleadings. It also raises the stakes for how courts prove up the record, verify facts, and communicate expectations to litigants who may never have had a lawyer. The numbers tell the story: a rising share of self represented filings, a doubling in volume since 2023, and a growing sense that AI is changing the pace and texture of civil litigation, even as winning chances for those litigants remain stubbornly flat.