Judges find suspected AI hallucinations in PA court cases
Judge Matthew Wolf identified numerous citation errors in an appellate brief, prompting review and potential sanctions amid rising concerns about artificial intelligence use in legal filings.
- On Dec. 10, Commonwealth Court heard oral argument after veteran attorneys filed an error-filled brief that raised questions about possible AI use, with Thomas Breth and Thomas W. King III presenting for the Thomas More Society.
- Researcher Damien Charlotin's database found AI hallucinations in at least 13 Pennsylvania filings, and Thomas Breth described the brief as collaborative, raising verification concerns about collaborative drafting practices.
- Commonwealth Court Judge Matthew Wolf interrupted to list citation problems, noting a quote attributed to the 2010 ruling Bayada Nurses, Inc. v. the Commonwealth of Pennsylvania, Department of Labor and Industry cannot be found in that opinion.
- The filings could trigger sanctions or fines from Commonwealth Court, and opposing counsel may refer the matter, potentially prompting investigations by the Office of disciplinary counsel and state Supreme Court disciplinary board.
- Experts warn the episode could erode attorney credibility and increase judicial skepticism, while the Pennsylvania Bar Association, U.S. Supreme Court Chief Justice John Roberts, and the American Bar Association say AI use is acceptable only with thorough fact-checking.
14 Articles
14 Articles
Judges find suspected AI hallucinations in PA court cases
Veteran attorneys with a track record of arguing high-profile cases submitted an error-filled brief to one of Pennsylvania’s appellate courts, raising questions from a judge about their use of artificial intelligence. When attorneys submit a brief with errors, whether AI-generated or otherwise, that is already a serious problem. But this happening in a case before Commonwealth Court is even more concerning, said David A. Harris, a professor at t…
Coverage Details
Bias Distribution
- 56% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium









