How AI may be American legal system's biggest new ‘problem’

Artificial intelligence tools such as ChatGPT and Claude are increasingly being used by people representing themselves in US courts. A report claims that this is contributing to a rise in filings that legal experts say can burden an already-stretched judicial system. While AI is helping some individuals prepare legal documents without hiring lawyers, courts are also seeing more cases built on weak arguments, inaccurate citations or large volumes of AI-generated paperwork.According to a New York Times report, a recent working paper by researchers Anand Shah and Joshua Levy, who examined more than 4.5 million non-prisoner federal civil cases and 46 million PACER docket entries, found that non-prisoner self-represented cases increased from a long-term average of around 11% to 16.8% in 2025.The study also reported that the number of docket entries generated by such cases in the first 180 days rose 158% above pre-AI averages by 2025, suggesting courts are handling more paperwork linked to self-filed lawsuits. Researchers found that, in a sample of 1,600 complaints filed between 2019 and 2026, more than 18% of those filed in 2026 were likely to contain AI-generated text.One example reported by the NYT involves Donald Sauve, a Minnesota resident who sued his former wife, her lawyer, and a state judge after an earlier challenge was dismissed. His first handwritten complaint seeking $275,000 in damages was rejected within weeks for lack of jurisdiction. Months later, Sauve submitted a revised complaint created with assistance from ChatGPT and Claude, alongside 50 additional filings, including what he described as a “case law synthesis.”The case was dismissed again, but not before court staff processed each filing. After repeated submissions, Judge Patrick J. Schiltz ordered that any further filings from Sauve would be “shredded without any additional notice.” Cases filed pro se, that is, without legal representation, have long been seen in the US courts. Many involve prisoners challenging their convictions or prison conditions.But worries about AI are increasingly focused on non-prisoner litigants who could use chatbots to generate legal documents that look professionally written. Supporters say AI tools could help people who can’t afford legal representation to navigate a system often viewed as expensive and difficult to understand. Earlier this year, Judge Michael Y. Scudder wrote that the technology offers “great promise” for improving access to justice.At the same time, courts have stressed that litigants remain responsible for inaccuracies in AI-assisted filings. In January, the Seventh Circuit warned that standards regarding accuracy and honesty still apply, even after a self-represented filing reportedly included AI-generated false citations.Concerns over fabricated legal references are not new. Previous cases involving lawyers using AI tools revealed fake precedents and invented court decisions. What appears to be changing is the scale, as access to generative AI expands beyond legal professionals to the general public.The growing use of AI in legal filings has created a new tension for courts: balancing wider access to legal resources with the challenge of filtering inaccurate or unsupported claims before they consume judicial time and resources.