November 20, 2026


Delaware Ruling Shows Why Law Firms Need More Than an AI Policy

A Delaware court’s decision to sanction a lawyer over fabricated legal citations while sparing his firm has underscored a growing challenge for law practices: adopting rules for artificial intelligence is only the beginning.

In a November 13, 2026, order, the Delaware Court of Chancery found that the firm had made substantial efforts to prevent the kind of misuse that led to the flawed filing. Those efforts, the court concluded, amounted to exceptional circumstances that relieved the firm of joint responsibility for the lawyer’s violation.

The lawyer had submitted a brief containing fabricated quotations and mischaracterized legal authorities. After opposing counsel raised the problem, he filed a corrected brief that removed quotation marks but retained inaccurate legal propositions, according to the source material.

The lawyer blamed a paralegal and argued that opposing counsel should have attempted to resolve the matter before bringing it to the court. The judge rejected that argument, writing, “There is nothing to negotiate or compromise when an attorney submits false citations to a tribunal.”

The court imposed sanctions under Delaware’s version of Rule 11 and required the lawyer to pay the monetary sanctions personally. The rule generally holds a law firm jointly responsible for a violation by a partner, associate or employee, absent exceptional circumstances.

In assessing the firm’s conduct, the court also considered Delaware’s professional conduct rules and ABA Formal Opinion 512, which addresses lawyers’ use of generative AI. It found that the firm had taken a series of steps to manage the risks.

The firm barred unapproved generative AI tools in March 2023, formed a working group in 2024 to evaluate tools and ethical guidance, and adopted a comprehensive policy in July 2025. The policy required an experienced lawyer to review AI-generated output for accuracy and completeness.

The firm also trained personnel on hallucinations and professional obligations involving competence, confidentiality, candor and supervision. Delaware’s Commission on Continuing Legal Education accredited the program for an hour of enhanced ethics credit.

The ruling illustrates why effective AI governance requires more than a written policy. Firms need to identify approved tools, protect confidential information and require verification of citations and other output. Training matters, too: lawyers and staff need to understand both the rules and the reasons behind them, particularly when deadlines make unapproved tools tempting.

Policies must also keep pace with developments such as AI agents. That requires firms to monitor changing technology, risks and legal guidance—and to have a practical way to approve revisions without lengthy delays.

The Delaware firm’s precautions did not prevent an individual lawyer from bypassing them. But they helped persuade the court that the firm had acted reasonably to prevent the misuse.