More Lawsuits… Thanks AI

A recent study released March 20 notes that pro se plaintiff filings in U.S. federal courts have nearly doubled since the arrival of generative AI. Perhaps even more interesting—intra-case docket activity is up 158%. In short, there are more cases and more activity within those cases.

Key takeaways come from Access to Justice in the Age of AI: Evidence from U.S. Federal Courts by Anand Shah and Joshua Levy. Their review presents evidence that AI is enabling lawsuits that, prior to its availability, would not have been filed, and that it is also increasing activity within lawsuits.

The Numbers: A Surge in Pro Se (Non-Criminal) Plaintiff Filings

Short version: For the 20-year period (2005–2025), FY 2025 had the highest number of total filings and the highest number of pro se filings, outpacing even the wave of lawsuits in the wake of the 2008 financial collapse.

A key observation is a dramatic increase in self-represented (pro se) plaintiff filings beginning in fiscal year 2023—the period immediately following the public release of large language model tools. Pro se filings jumped from a long-term pre-AI average to 27,370 in FY2023, then to 31,478 in FY2024, and 41,490 in FY2025. As the authors report, “The FY2025 count is almost double the pre-AI mean.”

This is not a minor fluctuation. Using Administrative Office of the U.S. Courts filing data, the authors calculate that “by FY2025, total civil litigation had increased by roughly 31,170 cases relative to the pre-AI mean (a 14.4% increase). Pro se filings alone accounted for 18,280 of those additional cases… new pro se filings account for 59% of the growth in civil filings.”

In other words, total civil litigation is growing significantly, and most of the growth is due to new pro se cases.

More Activity Within Each Case, Not Faster Resolution

Perhaps equally important is what is happening within filed cases. The Shah and Levy study finds that “intra-case activity is up, with the total volume of docket entries per court generated by pro se cases in their first 180 days up 158% from pre-AI levels to 2025.”

The study further concludes that “cases are not resolving any faster, and analysis of case disposition suggests little change in the quality of cases brought by pro se litigants. However, more is happening within each case.”

What’s Going On?

Fewer barriers to entry. Less cooling-off time.

Moving from perceived injustice to a filed lawsuit previously required a discussion with legal counsel—a gatekeeper who would assess the lawsuit’s potential value and translate the issues into an initial complaint. Both of those steps took time: schedule the initial consultation, have a meeting, receive the follow-up email with questions, draft the complaint, review, and file. Several days at the very least.

Now an initial complaint may be only hours away. AI allows an aggrieved person to prepare a complaint much faster and assess related jurisdictional and filing questions, without needing to either pay for or convince an attorney that the claims are worthwhile.

Quality Is a Related Issue

Already, there are thousands of documented material AI errors in court filings. One researcher, Damien Charlotin at HEC Paris, has set up an “AI Hallucination Cases” database tracking AI fabrications, false quotes, and case law misrepresentations. There are multiple entries every single day.

Those are just the publicly documented cases where a court goes through the mechanics of identifying and responding to AI errors. The full universe of AI errors entering the court system is staggering to consider.

Low-quality legal work is also becoming harder to spot quickly. AI programs are very good at producing neat, properly spaced, convincing filings with pristine Bluebook citations. The errors tend to be in the substance—cited cases that never existed or that stand for the opposite proposition.

What It Means—And What We Don’t Know Yet

We’re watching a natural experiment unfold in real time. The surge in filings and activity is undeniable, but the ultimate significance of this trend depends on how these cases resolve over the coming months and years.

The short version is that there are far more lawsuits—and more discovery, interrogatories, and related activity—to manage.

My prediction is that, in the short term, there will be a growing flood of legal claims, and discovery tools will be abused. 

Long term, the result may be so painful and expensive for enough people and businesses that meaningful litigation reform and alternative dispute resolution become political priorities. Over the next few years, I expect many more individuals and mid-sized businesses will come to realize how much power state district and superior court systems have to impact their day-to-day lives.