Courts Wrestle AI Fueled Lawsuits and Self Representation

Artificial intelligence is fueling a surge of lawsuits filed by people without lawyers.

A sweeping study of 4.5 million federal civil cases from 2005 through 2026 shows a sharp rise in self represented litigants, increasing from 11 percent in 2022 to 16.8 percent in 2025, with filings by nonlawyers more than doubling since 2023. The paper indicates that AI is part of this rising tide, as judges increasingly rely on AI tools to sift through pleadings and identify patterns that humans might miss. In Colorado, federal magistrate Judge Maritza Braswell is among those watching the trend up close. She uses AI to vet court documents and has said she can tell when AI generated prose or even fabricated quotes appear in filings. Her assessment of the practical impact is mixed: AI can produce better drafted pleadings, but it does not necessarily improve the chances of a plaintiff winning, and it raises new questions about the responsibility of chatbots in legal advice.

The real story, however, is not a single courtroom anomaly but a policy and practice shift. With more people turning to AI assisted tools, courts are learning to distinguish human from machine writing and to decide what kind of accountability, if any, should attach to the software. Judges are asking tough questions about whether a chatbot should bear duties akin to counsel, and lawmakers across the United States are grappling with who pays when AI doles out bad legal guidance. The tension is clear: AI can improve access to justice by lowering the bar to filing, but it can also flood the system with imperfect, sometimes misleading, pleadings.

For engineers and product leaders, the implications are concrete. First, there is a clear demand for reliable “human in the loop” workflows. Relying on AI to draft or review filings must be paired with human verification, ideally integrated into case management dashboards that flag AI generated content, hallucinations, or quotes that do not check out. Second, governance matters. The courts’ experience suggests that model behavior should be auditable, with outputs traceable to sources and prompts, an essential step if AI becomes an everyday tool in legal workflows. Third, the incentives around AI use in justice must be aligned. Access to justice is improved only if the quality of representation and information does not degrade into misdirection or misinterpretation, a risk that grows with scale and with a heterogeneous user base of laypeople.

As the trial bar and policy makers weigh guardrails, one practical takeaway is that the telltale signs of AI assistance, namely stylized prose, plausible but incorrect quotes, and inconsistent citations, require explicit human review. Judges like Braswell are already building the playbooks for what to look for, and lawmakers are debating who should bear the costs when bot advice goes awry. The next phase will likely hinge on standards for AI assisted pleadings, explicit disclosure of tooling, and clearer accountability rules that keep the door open for accessible legal help without inviting a flood of unreliable filings.

What to watch next: court guidance on AI disclaimers and attorney oversight, the emergence of trustworthy AI audits in judiciary workflows, and state or federal steps to budget for AI literacy and enforcement of reasonable standards in AI generated pleadings.