The growing use of artificial intelligence in legal work could create a new problem for the profession: junior lawyers may lose opportunities to develop the skills they traditionally acquire by drafting and checking legal documents themselves.
US District Judge Arun Subramanian of Manhattan raised that concern while declining to penalize lawyers whose court filing contained citation errors linked to the use of AI.
In an order issued this week, Subramanian described the episode as a warning for the legal profession and stressed that material generated by AI must be independently verified before it reaches a court.
The judge also pointed to a broader concern for clients. While some clients may want law firms to adopt AI to reduce costs and improve efficiency, mistakes generated by such systems can have consequences that outweigh those savings.
Subramanian said clients may therefore want to consider carefully whether they should require their lawyers to rely on AI for work such as preparing legal briefs.
The dispute arose in a copyright case in New York. Subramanian had previously questioned whether unchecked reliance on AI contributed to inaccurate citations in a filing submitted on behalf of Foundation Media.
The court ultimately decided against imposing sanctions.
Attorney Cynthia Arato said the problematic filing had been prepared and finalized by another lawyer at her firm. According to a declaration submitted to the court, that lawyer had acted against the firm’s established policies and training requirements.
The errors, Arato said, resulted from a failure to verify AI-generated citations. She characterized the incident as isolated and said there was no intention to mislead the court.
A subsequent filing revealed that the firm, Shapiro Arato Bach, had used AI tools developed by Harvey while preparing the filing in question. The firm said it was putting additional safeguards in place to prevent similar problems.
The episode also prompted Subramanian to focus on an issue extending beyond inaccurate citations: how young lawyers will learn their profession if increasingly sophisticated AI systems perform much of the basic drafting work.
He urged the legal industry to consider ways of ensuring junior attorneys continue receiving meaningful training as firms expand their use of AI.
One possibility, the judge suggested, would be for firms to require lawyers to prepare initial drafts of briefs without AI assistance. Such a process could preserve opportunities for junior attorneys to learn legal research, analysis and drafting before technology is brought into the workflow.
Subramanian acknowledged that AI may become an increasingly unavoidable part of legal practice, but emphasized that developing the next generation of lawyers remains a shared responsibility.
The warning comes as courts across the United States continue to encounter filings containing inaccurate or nonexistent citations and other errors associated with AI-assisted legal work. Judges are increasingly being forced to determine how attorneys should be held accountable when technology contributes to mistakes in court submissions.
For law firms, the issue is consequently becoming larger than whether AI can make legal work faster. It also involves deciding how much of the profession’s traditional hands-on training can be delegated to machines without weakening the skills lawyers ultimately need to protect their clients.
