Pro se Filings Rise as AI-Generated Complaints Appear
Per a draft paper by Anand V. Shah and Joshua Y. Levy (hosted on GitHub and excerpted by Reason), the share of federal civil cases filed pro se rose from a long-term average of 11% to 16.8% in FY2025. The authors report the increase is concentrated in case types characterized by formulaic document production and absent from more complex, attorney-intensive categories. The paper also reports that pro se-driven docket activity per court in the first 180 days rose 158% relative to pre-AI means through 2025, increasing judges' case-processing workload. In a random sample of 1,600 complaints from 2019-2026 the authors find AI-generated text flags grew from near zero to over 18% in 2026. Editorial analysis: This trend highlights growing intersections between generative AI and access-to-justice dynamics, with implications for legal workflow tooling and document-forensics demand.
What happened
Per the draft paper by Anand V. Shah and Joshua Y. Levy, hosted on GitHub and excerpted by Reason, researchers analysed administrative records covering more than 4.5 million non-prisoner federal civil cases (FY2005-FY2026) and 46 million PACER docket entries. The paper reports the share of pro se federal civil filings rose from a long-term steady-state average of 11% to 16.8% in FY2025. The authors report the increase is concentrated in case types described as formulaic and is absent from more complex, attorney-intensive categories. The paper further reports that total docket entries per court generated by pro se cases in their first 180 days rose 158% from pre-AI means to 2025. In a random sample of 1,600 complaints spanning 2019-2026 the authors find AI-text detection flags increased from essentially zero in the pre-AI period to more than 18% in 2026.
Editorial analysis - technical context
Observed patterns in similar datasets show that when document generation becomes template-like, generative models lower the production cost of filings. Industry-pattern observations: automated text generation is most likely to expand in high-volume, low-variation legal tasks, which can increase raw filing volume without commensurate increases in legal-service capacity. For practitioners: this raises demand for reliable AI-origin detection, robust document provenance tracking, and tooling to triage low-complexity filings.
Industry context
Industry observers have noted rising interest in legal-tech that automates form-filling and pleadings; the paper's findings align with that broader trend. Editorial analysis: increased pro se filings where AI assists drafting may create more friction for court clerks and judges because additional docket entries and procedural errors tend to follow self-represented filings, increasing administrative load even when filings are substantively simple.
What to watch
Editorial analysis: observers should track:
- •replication of these patterns in state courts
- •whether AI-detection methods used by the authors hold up across other samples and vendors
- •any operational changes in court rules or PACER workflows that respond to higher volumes of machine-assisted filings. The authors' dataset and code published on GitHub will allow practitioners and researchers to validate detection methodology and measure externality effects on court operations
Scoring Rationale
The paper documents measurable, system-level effects of generative AI on federal court filings and docket load, which matters to practitioners building legal workflows, detection tools, and data scientists monitoring downstream impacts. The story is notable but not a frontier-model release.
Practice interview problems based on real data
1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
