Attorney Maren Bam’s Rule 11 sanctions for recklessly filing AI-generated unverified citations in a Social Security case were largely vacated on appeal — because the sua sponte sanction process was not followed correctly. But her pro hac vice status in the case was revoked and her opening brief was struck, leaving real professional consequences.
An Australian federal court found that a self-represented bankruptcy applicant had relied on AI-generated legal authorities that did not exist — but showed leniency because the applicant was unrepresented, acknowledged AI use, and did not press the fictitious authorities when challenged. The court distinguished this from lawyer misconduct.
A federal patent court in Kansas ordered all signing attorneys for patent plaintiff Lexos Media to show cause why they should not be sanctioned and referred to disciplinary administrators for submitting AI-generated defective legal citations in both a summary judgment response and an expert witness brief.
A Michigan federal court remanded a Social Security disability case while warning plaintiff’s counsel that a cited case — Fleck v. Commissioner of Social Security — appeared to be a ‘phantom’ case likely generated by AI. The court warned that future AI-generated unconfirmed citations could result in sanctions.
A Massachusetts federal court referred attorney Roger Peace to the Massachusetts Board of Bar Overseers after he submitted AI-generated citations to nonexistent authority and gave an evasive response to the court’s show-cause order. The referral — rather than a direct sanction — represents a new mechanism for AI citation accountability.
The Pennsylvania Superior Court published a decision in a car loan dispute where court submissions contained AI/LLM-generated citations to cases that do not exist. The Westlaw annotation explicitly identifies large language model (LLM) hallucinations as the source, marking one of the first state appellate courts to use that specific terminology.
A Florida federal court held a sanctions hearing over AI-generated fake citations submitted in two consolidated employment cases, warning that ‘the use of AI as a legal research tool does not alter’ lawyers’ basic ethical obligation to provide true and accurate citations to the court.
A Chicago pro se plaintiff brought Fair Housing Act claims against his landlords, but his filings contained references to cases that do not exist. The N.D. Illinois court granted dismissal of some claims while flagging the hallucinated citations in its published opinion.
A federal court in Michigan denied a motion for relief from judgment in a case where Westlaw editors flagged the filed submissions as containing non-existent citations generated by ChatGPT and OpenAI tools. The case is another data point in the escalating judicial reckoning with generative AI in legal practice.
The Appellate Court of Maryland flagged non-existent case citations in a divorce appeal brief — citations that did not exist and were likely AI-generated. The Westlaw editor’s note explicitly identifies the decision as containing ‘incorrect or non-existent’ references, a pattern courts are increasingly calling out.