Long before large language models could generate a legal brief at the click of a button, legal scholars warned that democratizing access to litigation tools would fundamentally reshape the judiciary. Today, those warnings have materialized in a striking new form: courts across the country are being inundated with lawsuits crafted entirely — or largely — by artificial intelligence systems wielded by self-represented plaintiffs.
The phenomenon traces a direct line back to the early desktop publishing revolution of the 1980s, when typewriters gave way to word processors and the volume of pro se filings began a steady climb. Each successive wave of accessible technology — from legal self-help websites in the 1990s to document-automation software in the 2000s — nudged that number higher. Generative AI represents the steepest jump yet, effectively placing a tireless, if sometimes hallucination-prone, legal drafter in anyone's pocket.
Federal judges have already begun encountering briefs that cite nonexistent case law — a signature flaw of AI-generated text — while court clerks report processing filings that are structurally coherent yet legally frivolous. The pattern echoes a concern computer scientists raised as far back as the 1970s during early natural-language processing research: that a system fluent in the form of expert discourse might still be wholly ignorant of its substance.
What distinguishes this moment historically is scale. Individual bad actors have always flooded dockets with nuisance suits, but AI lowers the marginal cost of filing to near zero, meaning the judiciary must now contend with industrial-scale amateur litigation rather than isolated incidents. Some legal reformers argue the solution lies in updated filing fees or mandatory disclosure requirements for AI-assisted documents. Others see it as the inevitable friction of a society still calibrating the boundary between technological empowerment and institutional harm.
Either way, courts built for a pre-digital age are now serving as the unlikely proving ground for one of the central questions of the AI era: who bears responsibility when a machine helps an untrained user navigate — or overwhelm — systems designed by and for professionals?