Federal judge links AI quality control to junior-lawyer development

On October 2, 2026, Reuters reported that U.S. District Judge Arun Subramanian declined sanctions over AI-generated citation errors but called the episode a warning about unchecked use and weakened junior-lawyer training. The responsible lawyer had acted contrary to firm policies and training, and the firm said it was adding safeguards; the judge suggested initial brief drafts could sometimes be prepared without AI. For mid-sized firms, a written policy is therefore insufficient unless matter supervision, citation checks and exception handling can be demonstrated in practice. Leaders should also treat training design as part of AI governance, preserving the work through which younger lawyers learn legal reasoning while automating lower-value steps around it.