Lawyers worldwide are increasingly filing court documents containing fake legal precedents generated by artificial intelligence. Recent data compiled by HEC Paris tracks over 1,200 instances of AI-driven hallucinations in legal filings, with 800 of those cases occurring in the United States alone.
Despite high-profile sanctions and six-figure fines from courts, the rate of these filings continues to climb. The trend has persisted even after initial public warnings in 2023, when lawyers in New York first faced consequences for submitting AI-authored briefs that cited non-existent cases.
The cost of automated shortcuts
Legal experts point to a lethal combination of efficiency pressure and the tendency of large language models to fabricate facts that appear credible. While AI can produce structured documents quickly, these systems frequently generate citations that do not exist, misleading courts and undermining judicial integrity.
Some firms have exacerbated the problem by forcing junior staff to use AI tools without providing access to the verified legal databases necessary for fact-checking. This exploitative workflow prioritizes speed over accuracy, leading to a surge in AI errors that now spans at least ten different jurisdictions.
Responsible practitioners argue that AI tools only provide value if they are used judiciously and paired with extensive manual verification. These lawyers report that the time required to cross-reference AI output often matches the time saved during the initial drafting process.
While industry proponents suggest that AI may eventually improve through better infrastructure, current models show no sign of eliminating hallucinations. The ongoing reliance on these tools suggests that, for many in the legal sector, the lure of productivity gains currently overrides the risk of professional censure.
