The American legal system is facing an unprecedented crisis as federal courts are inundated with a flood of self-filed lawsuits generated by artificial intelligence tools like ChatGPT and Claude. What began as a trickle of handwritten complaints from individuals seeking justice without legal representation has swelled into a torrent of AI-crafted legal documents that threaten to overwhelm the judiciary’s capacity to function. The convergence of accessible AI language models and a growing population of pro se litigants has created a perfect storm, raising existential questions about the sustainability of the federal court system and the very nature of access to justice in the digital age.
The Surge of AI-Powered Pro Se Litigation
The phenomenon of pro se litigation—where individuals represent themselves without an attorney—is nothing new in American jurisprudence. However, the scale and sophistication of such filings have been transformed by the emergence of powerful AI tools. These technologies enable individuals with limited legal knowledge to produce documents that appear professionally crafted, complete with citations, legal arguments, and formal structure. The result has been a dramatic increase in the volume of cases filed in federal district courts, which already handle approximately 300,000 civil cases annually.
Donald Sauve: A Case Study
The story of Donald Sauve, a 69-year-old Minnesota resident now living out of his car in Mankato, illustrates both the promise and the peril of AI-assisted litigation. Sauve initially filed a handwritten lawsuit against his ex-wife, her attorney, and the state judge who had dismissed his earlier case as frivolous. When his handwritten complaint seeking $275,000 in damages was dismissed within a month for lack of jurisdiction, Sauve returned with a radically different approach. Using ChatGPT and Claude, he submitted a professionally formatted complaint accompanied by fifty additional filings, including an integrated analysis of case law. The result was the same—dismissal—but not before court clerks had to read, organize, and docket hundreds of pages of material.
Chief Judge Patrick J. Schiltz of the Minnesota federal district court issued a 14-page opinion dismissing the case and ordered that any future filings be discarded without notice. As he wrote, courts cannot be expected to sift through hundreds of pages of documents to extract facts and arguments. Sauve, however, remains undeterred, planning to submit what he calls Supreme Court-level filings to state and federal appellate courts, convinced that the law is on his side because OpenAI and Claude told him so.
Statistical Evidence of the Crisis
The scale of the problem is now measurable thanks to research from MIT and USC. A study by Anand V. Shah and colleagues examined 1,600 randomly sampled complaints filed between 2019 and 2026, finding that the proportion containing AI-generated text surged from virtually zero to over 18 percent. Among non-incarcerated pro se litigants, the share of self-filed cases jumped from 11 percent five years ago to 16.8 percent in 2025. In the Minnesota federal district alone, pro se filings by non-prisoners increased by approximately 50 percent since March 2025, spanning claims from false arrest and malicious prosecution to divorce disputes and even cases alleging that all eviction lawsuits violate fair housing laws.
MIT Study Reveals Alarming Trends
The research paints a clear picture of exponential growth. AI-generated text in federal court filings increased from negligible levels in 2023 to 3.5 percent in 2024, then to 10.5 percent in 2025, and is projected to exceed 18 percent in 2026. These numbers represent not just statistical curiosities but real burdens on a system already operating at capacity. Each AI-generated filing must be reviewed, categorized, and responded to, consuming judicial resources that could otherwise be directed toward meritorious claims.
The Judicial Burden: An Existential Threat
Federal judges across the country have begun to speak out about the strain. Judge Schiltz, who oversees the pro se docket in Minnesota, described the situation as an existential crisis for the federal courts, adding that no end is in sight and no satisfactory solution is apparent. His sentiment is echoed by colleagues nationwide who report spending increasing portions of their dockets on AI-generated filings that lack legal merit but require the same procedural attention as legitimate cases.
Judge Schiltz’s Warning
The burden is not merely quantitative but qualitative. AI-generated filings often exhibit what experts call a hallucination problem—the creation of seemingly plausible but entirely fabricated information. In the legal context, this manifests as citations to nonexistent cases, invented statutes, and arguments built on fictional precedents. Judge Joshua D. Wolson of the Eastern District of Pennsylvania noted at a May judicial conference that the cost of producing things that look real but are not real is going down and quality is going up, creating a profound problem for the courts. Judge Virginia Kendall of Illinois imposed a $1,500 sanction on a litigant who twice submitted filings containing fabricated cases, stating that time of both parties and the court was wasted chasing down phantom precedents.
The Double-Edged Sword of AI Access
The debate over AI in pro se litigation reveals a fundamental tension in American legal philosophy. On one hand, AI tools offer unprecedented access to justice for individuals who cannot afford attorneys. Judge Michael Y. Scudder of the Seventh Circuit Court of Appeals wrote in a pro se decision that AI holds significant potential to improve judicial access for people who cannot afford to hire a lawyer. Steven Donohue, who manages the pro se docket in Minnesota, acknowledged that when used properly, AI could be an extremely powerful tool for those who believe they have been treated unfairly.
Democratization vs. Dysfunction
Historically, pro se litigation has driven important legal reforms. In 1963, Clarence Earl Gideon’s handwritten petition to the Supreme Court established the right to counsel for felony defendants. AI could amplify such voices, enabling individuals with legitimate grievances to articulate their claims effectively. Yet the current reality is that most AI-generated filings are formulaic, legally deficient, and burden the system without advancing justice. The technology’s ability to mass-produce documents that look real but lack substantive legal grounding threatens to overwhelm the very institutions designed to hear those voices.
Legal and Regulatory Responses
Courts are beginning to fight back, but their options are limited. The right to file a lawsuit is constitutionally protected, and there are few mechanisms to restrict access to the courts based on the tools used to prepare filings. Some district courts have issued standing orders warning about the risks of using AI without professional oversight. At the state level, legislators are exploring bills that would hold AI companies responsible when their chatbots provide legal advice that leads to frivolous litigation, though no such measures have been enacted.
The Administrative Office of the U.S. Courts acknowledges the problem, stating it is aware of this issue and is gathering information about its impact on the system. Both OpenAI and Anthropic include in their terms of service that output responsibility lies with the user and that their tools should not be used as a substitute for professional advice. Such disclaimers, however, are invisible to users like Sauve, who trust the AI’s assurances over corporate legalese.
The Road Ahead: Between Access and Overload
The question facing the American judicial system is whether AI represents a democratization of justice or a pathway to institutional dysfunction. The technology is not going away, and its capabilities will only improve. Courts must adapt by developing efficient screening mechanisms for AI-generated filings, while preserving access for legitimate pro se litigants. Some have suggested specialized pro se dockets, AI-assisted triage systems, or mandatory certification requirements for AI-generated legal documents. The answer will ultimately be determined by the courts themselves, already operating at the breaking point. As AI opens doors that were previously closed to those without legal representation, the institution tasked with upholding justice must decide whether those doors lead to a more equitable system or to its own collapse. The outcome will redefine the relationship between technology and the law for generations to come.
