Courts grapple with AI driven flood of self represented lawsuits

AI is fueling a flood of self represented lawsuits in federal courts.

A new study tracking 4.5 million federal civil cases from 2005 to 2026 shows the share of filings by people without lawyers jumped from 11% in 2022 to 16.8% in 2025, and filings in this space more than doubled after 2023. The authors tested whether AI played a role in that surge, and the signal is hard to ignore for judges who see the docket from both sides of the aisle. In Colorado, Judge Maritza Braswell notes a rise in pro se filings that she links, at least in part, to the spread of AI tools. She uses AI herself to vet court documents and to screen filings written with little legal guidance. She has learned to spot AI style in the prose and even the odd hallucinated quote, yet she also observes that pleadings drafted with AI can be sharper and more complete than those produced by some unrepresented litigants.

The paper shows a troubling paradox: AI may lower the barrier to entry for people who cannot afford counsel, but it does not appear to tilt outcomes in favor of those litigants. Judges report that while AI can help with organization and language, it does not automatically translate into better chances of winning. The study’s authors, Anand Shah and Joshua Levy, say they tested whether AI involvement correlates with more filings by self-represented litigants. The answer is a correlation that aligns with what Braswell sees in real time: more AI assisted or AI generated pleadings coincide with more pro se filings, but without a clear uplift in success rates for those filers.

Beyond the courtroom floor, the questions are shifting from capability to responsibility. Lawmakers are weighing who should bear the cost when chatbots give bad legal advice and what duties a chatbot should owe in high stakes settings. The arena is moving from technologist bragging rights to accountability questions on misrepresentation, due process, and the line between tool and legal advisor.

From a practitioner perspective, a few clear takeaways emerge. First, courts are being forced to design guardrails around AI assisted filings. Braswell’s experience suggests judges will need to improve detection of AI styled text and hallucinated quotes, while still recognizing when AI has produced meaningful, well-structured pleadings. That implies new workflow norms: faster triage for AI aided submissions, paired with human in the loop review to catch factual gaps or invented quotes before they reach a courtroom.

Second, the on ramp for pro se litigants is real but uneven. Access to justice improves when AI helps draft or organize a case, but the quality of that output varies wildly depending on user expertise and the availability of reliable prompts or templates. Third, accountability and funding are rising up the policy agenda. If AI generated advice contributes to losses or procedural missteps, who pays and who is responsible becomes an open question for courts and legislators alike. Finally, the practical engineering constraint is clear: tools that help with form and prose must be coupled with verification and auditing mechanisms that prevent hallucinations and misstatements from guiding real legal outcomes.

In short, what was once a quiet AI experiment in the background is now a courtroom reality. The courts are adapting, and so are the people who rely on them. The coming months will reveal whether guardrails keep pace with accessibility, and whether the justice system can harness AI to help the underserved without letting them down in the courtroom.