Swiss Courts Report a Surge of Lengthy ChatGPT-Written Submissions From Litigants

Courts across Switzerland are receiving a growing volume of long, AI-drafted legal submissions, especially from people representing themselves, according to a survey of several courts by 20 Minuten published on 16 October. Judges say the texts often sound polished but drift off the point, cite rulings that have nothing to do with the case and add work to proceedings that are already stretched.

The Zurich District Court said it could not yet see AI driving more cases, but legal laypeople were increasingly filing apparently AI-generated texts that were linguistically correct yet overly long, repetitive or poorly tailored to the case, sometimes with inaccurate references to case law. Processing effort can rise considerably, it said, particularly in time-critical family-law matters where extra urgent motions must be checked quickly. Courts in Bern, Aarau and Lucerne reported similar patterns.

In Fribourg, the judicial council had already spoken this spring of structural overload, with new cases up about 16 percent since 2021, although that rise cannot be pinned on AI. At the Basel-Stadt civil court, president Patrik Müller said laypeople have turned to AI over roughly the past year and that pages-long statements now arrive even in simple cases. With about 6,000 proceedings a year, he warned, the court’s ability to decide within a reasonable time is at risk.

What it means in practice

Müller does not think banning AI in proceedings is realistic. He wants courts to get AI assistance that captures filings and summarises what is actually relevant, while the decision itself stays with a human judge. Klaus Vogel, president of the Swiss Association of Judges, said longer filings mean longer processing times but called warnings of a blocked justice system overstated. The debate runs alongside the early-2027 consultation draft for a Swiss AI law and the Federal Council’s five priority areas for AI in the federal administration.

There are upsides: courts note fewer confused or insulting letters and better participation by overwhelmed parties. But Müller added that people advised by a chatbot tend to trust it more than the court, because the tools are built to affirm their users, unlike a lawyer who tells a client when they are on the wrong track.

What to watch: cantonal guidance on AI-drafted filings, data-protection decisions on court-side AI tools, and budget requests for more court staff if the trend continues.