The scholar is not named in the Law360 excerpt reviewed, so the attribution rests on JD Journal alone. JD Journal also states that the trend does not prove AI errors are becoming less common overall. Before the new data, LawDistrict had projected roughly 1,200 to 1,300 documented cases by the end of 2026.
The errors are also changing. One analysis of the database found that 53% involved fabricated citations to cases that do not exist, 26% misrepresented real cases, and 20% involved false quotations. The scholar argues that litigation risk now extends beyond fabricated authority, so verification duties may need to cover holdings, descriptions, and quotations, not just whether a case exists. Courts are still pressing the point: the US Tax Court flagged fake and AI-garbled citations in two 2026 opinions, one involving a self-represented taxpayer. JD Journal reports that false claims and weak arguments waste court time. Attorneys should check every citation and quotation in AI-assisted filings against the source before filing.