Courts grapple with AI fueled flood of filings

Self-represented filings jumped to 16.8% in 2025 as AI assisted filings flood courts. A new study surveying 4.5 million federal civil cases from 2005 to 2026 shows the rise in litigants without lawyers coinciding with a surge in AI-enabled submissions. The paper shows the share of self-represented cases rising from 11% in 2022 to 16.8% in 2025, and filings more than doubling compared with pre-2023 levels. The numbers illuminate a legal ecosystem where AI is shaping both access to justice and the day-to-day grind of the judiciary.

In Colorado, federal magistrate judge Maritza Braswell is among the many judges adapting on the fly. She uses AI tools to vet court documents, and she notes she can spot when large language models have written a filing, often by the prose and by occasional hallucinated quotes. "I correlate that to AI in part because I see AI use," she says. The trend her courtroom is facing is double-edged. AI can help people who lack counsel articulate their arguments more clearly, but it can also flood dockets with submissions that require careful scrutiny. In Braswell’s experience, the surge has added to workloads even as the pleadings themselves become easier to understand and more coherent. The team reports that, on balance, AI-assisted filings appear to improve drafting quality for self-represented litigants, even as the court sifts through the human and machine-made text.

The broader implications extend far beyond a single courtroom. The study's trend lines suggest AI is enabling greater access to the legal process, but judges are wrestling with what responsibilities, if any, should attach to the AI outputs themselves. The paper shows that while AI seems to help people articulate their arguments, it does not automatically improve their odds of winning, a reality that sits at the intersection of accessibility and accuracy. Benchmarks indicate a growing need for guardrails around AI generated content, including how to verify citations, ensure jurisdictional relevance, and prevent the spread of quotes or cases that do not exist. The team reports that courts are already encountering cases where AI-aided filings bear plausible-sounding but fabricated authority, underscoring the risk of overreliance on machine-generated precision.

Policy conversations are heating up at state and federal levels about who pays when AI misbehaves in a legal setting. Lawmakers across the United States are grappling with accountability, licensing, and liability for chatbots acting in quasi-legal roles. The central engineering constraint is clear: AI is a powerful tool for widening access, but it requires robust human oversight, transparent disclosure of AI use, and reliable verification workflows to keep pace with evolving filing patterns. For practitioners and product teams building tools for law offices or court staff, the takeaway is equally pragmatic. First, treat AI as a drafting and triage aid, never as a stand-in for legal judgment. Second, design interfaces that flag potential errors, cite sources clearly, and require human review before submissions. Third, anticipate docket pressure: AI can accelerate entry of papers, but it also compounds the need for line-by-line validation, cross-checking, and risk assessment. Finally, monitor regulatory and funding shifts; the AI-enabled courtroom is still a developing ecosystem with evolving rules on responsibility and cost.

As courts adjust to this new normal, the tension between expanding access and maintaining reliability will shape tools, procedures, and policy for years to come.