Unless firms replace lost feedback, AI may weaken lawyer training

A federal judge warned that law firms’ increasing use of artificial intelligence could create risks for clients and hinder the training of junior lawyers, Reuters Legal reported on October 2. The judge declined to sanction attorneys over AI-generated errors in a court filing. The immediate event therefore has two parts: the court did not impose a penalty, but it identified client risk and junior training as consequences of how firms use AI.

That distinction matters for AI-native law. The absence of a sanction is not the absence of accountability; it is a reminder that legal work is judged through more than the punishment attached to a particular filing. Legaltech News says AI supplies speed, scale, consistency and computational power, while attorneys supply legal knowledge, context, skepticism, judgment and accountability. The judge’s warning fits that division: AI can increase the production of legal material without taking responsibility for whether the material is fit to reach a client or a court. The consequence arrives through supervision, professional judgment and the quality of the work, not only through a fine.

The training point is equally important. Legaltech News reports that artificial intelligence is taking the lead on research, drafting and document review while firms struggle with how junior lawyers can develop skills traditionally learned through repetitive work. That creates a real contradiction for firms that automate the work used to teach judgment while depending on experienced lawyers to supervise the output. Automation can remove bad or low-value work, but it does not by itself create the feedback that replaces the lost practice. The American Lawyer makes the same distinction more directly: AI is not the threat to lawyer training; lost feedback is. A firm can therefore be efficient at producing drafts and still fail to produce its next generation of capable lawyers.

The event proves a narrower point than a general prediction about the profession. It establishes that a federal judge warned about client risk and junior-lawyer training while declining sanctions over AI-generated errors in a court filing. It does not establish that AI routinely produces defective legal work, that every firm has lost its training pipeline, or that leaner staffing is economically superior. A stronger reading would require firms to show that AI use lowers the quality of supervision or removes meaningful feedback, alongside a measured change in lawyer development or work quality. The warning makes that question concrete; it does not answer it. Nor does it turn a court’s decision not to sanction attorneys into approval of unsupervised AI use.

The next observable test is an order or filing that explains what review the attorneys performed, what AI-generated material entered the submission, and why the court treated the errors as insufficient for sanctions. Firms’ training disclosures are the parallel test: whether they report structured feedback, lawyer review and a path for juniors to acquire judgment after AI takes over research, drafting and document review. Privilege handling supplies another boundary. Legaltech News reports that AI agents disrupt the traditional workflow in which an attorney inputs information, software processes it and the attorney reviews the output, with privilege and confidentiality obligations running through the attorney’s conduct. We should count the model as viable only when the next filing and the next training numbers show both accountable review and a functioning route to develop lawyers.