A US federal judge has declined to sanction lawyers over AI-assisted citation errors—but warned that increasing reliance on legal AI could harm clients and weaken the training of junior practitioners.
The errors arose in Hill v Foundation Media LLC, a copyright proceeding in the Southern District of New York. A partner at Shapiro Arato Bach apologised and disclosed that Harvey had assisted with the inadequately verified filing. She said another lawyer acted contrary to the firm’s policies and training. Judge Arun Subramanian accepted that the incident did not involve bad faith, while describing it as a “wakeup call” for the profession.
The judge suggested firms consider requiring junior lawyers to prepare initial drafts without AI. His concern was not simply accuracy: if lawyers outsource the work through which they learn to research, analyse and write, future clients may receive poorer professional judgment.
That makes this development more consequential than another hallucinated-citation episode. Verification can catch an incorrect authority, but it cannot replace the expertise that lawyers develop by constructing arguments themselves.
For Australian firms, the practical lesson is that AI governance should protect professional formation as well as work product. Adoption policies may need to identify tasks lawyers must first perform independently, alongside requirements for source checking, supervision and final practitioner responsibility.
Sources: Reuters; case docket
Leave a Reply