Artificial IntelligenceLawyerTechnology

Delaware Lawyer Fined Over AI Errors, Firm Escapes Punishment

A Delaware attorney has been ordered to pay a $2,500 sanction after submitting a court filing containing fabricated quotations, inaccurate citations, and mischaracterized legal authorities produced with the assistance of generative artificial intelligence. The Delaware Court of Chancery held the individual lawyer responsible for failing to verify the accuracy of the material before it was filed. However, the court declined to punish his law firm, concluding that the organization had taken substantial steps to prevent the misuse of artificial intelligence through internal policies, restrictions, and mandatory employee training.

The decision was issued on November 13, 2026, by Vice Chancellor Lori W. Will in Kevin Leiske et al. v. Robert Gregory Kidd et al., a dispute involving requests for advancement of legal expenses connected to a broader investment-related conflict. The ruling addressed the conduct of Richard P. Rollo, a partner and director at Richards, Layton & Finger, who signed a legal brief submitted on behalf of the plaintiffs. The court had ordered both Rollo and his firm to explain why sanctions should not be imposed after serious inaccuracies were identified in the filing.

The controversy began with an answering brief submitted on January 22, 2026. Five days later, opposing counsel identified fabricated quotations, inaccurate citations, and legal propositions that did not accurately reflect the cited authorities. Rollo subsequently acknowledged that a generative AI tool had been used to revise the brief. A corrected version was submitted, but it did not resolve every problem. Although quotation marks around nonexistent quotations were removed, inaccurate descriptions of the underlying legal decisions remained in the document.

The filing errors raised concerns about the reliability of legal arguments presented to the court and the professional responsibilities of attorneys who use artificial intelligence. Rollo initially attributed the failure to identify certain errors to an oversight during a paralegal’s citation review. He also challenged opposing counsel’s decision to bring the inaccuracies to the court’s attention. However, he later accepted responsibility in a sworn affidavit and acknowledged that he had failed to properly verify the legal authorities included in the brief.

According to the court’s review of the circumstances, Rollo used Perplexity Enterprise to help draft and condense arguments. At the time, the software was being evaluated through a pilot process and had not been approved for general use by the firm. The AI-assisted drafting process introduced fabricated quotations, flawed citations, and an inaccurate description of a court decision into the filing. Rollo then delegated citation checking to a junior attorney and a paralegal before submitting the brief without personally confirming the accuracy of the authorities and quotations.

Vice Chancellor Will emphasized that the use of generative AI in legal practice is neither prohibited nor automatically improper. Artificial intelligence can provide lawyers with useful tools for research, drafting, and improving efficiency. Nevertheless, attorneys remain bound by professional standards requiring competence, accuracy, and honesty when communicating with courts. The judge stressed that lawyers cannot transfer their professional judgment or responsibility to an AI system, regardless of how advanced the technology may be or how many people participate in reviewing the resulting work.

Join YouTube banner

The court specifically relied on Court of Chancery Rule 11, which requires attorneys to conduct a reasonable inquiry before submitting court papers. The rule also reinforces the duty of candor under Delaware’s Rules of Professional Conduct. The judge explained that the obligation to verify legal authorities and ensure the accuracy of representations made to a court is personal and nondelegable. Having a junior lawyer or paralegal review an AI-assisted filing does not eliminate the signing attorney’s responsibility to ensure that the final document is accurate.

The court reached a different conclusion regarding Richards, Layton & Finger. Evidence showed that the firm had prohibited the use of unapproved generative AI tools as early as March 2023. In 2024, it established a working group of directors and professional staff to evaluate AI products and examine relevant ethical guidance. The firm’s board adopted a comprehensive generative AI policy in July 2025, requiring experienced attorneys to review AI-generated material for accuracy and completeness. The policy also prohibited using AI as a substitute for independent legal analysis and professional judgment.

The firm had also established an internal training program addressing AI hallucinations, confidentiality, professional competence, candor toward courts, and supervisory responsibilities. More than 200 employees had completed the training by January 2026, before the filing at issue was submitted. The program was accredited for enhanced ethics continuing legal education in Delaware. Vice Chancellor Will found that these measures demonstrated a serious institutional effort to prevent AI-related misconduct. Because the safeguards were already in place and Rollo had bypassed them, the court concluded that the firm should not be sanctioned for his individual failure.

In imposing the penalty, the judge considered Rollo’s acceptance of responsibility and expressions of remorse. The court also recognized his history of practice before the Delaware Court of Chancery and noted that he had completed multiple continuing legal education courses addressing AI ethics. Additional mandatory training was therefore considered unnecessary. Nevertheless, those mitigating circumstances did not eliminate the violation. The court ordered Rollo to personally pay $2,500 within 14 days of the November 13 decision and to submit an affidavit confirming payment. The firm was released from the sanctions proceeding without punishment.

The decision adds to growing judicial scrutiny of lawyers who rely on generative AI when preparing legal documents. It also offers an important distinction between individual professional responsibility and an organization’s efforts to establish reasonable safeguards. Law firms may reduce their exposure to sanctions by adopting clear AI policies, restricting unauthorized tools, and training employees on the technology’s risks. However, those measures cannot excuse an attorney who fails to personally verify the accuracy of a court filing. The ruling makes clear that AI can assist legal professionals, but responsibility for the arguments, quotations, and authorities they submit remains with the lawyers who sign their names to the documents.

Key Legal Outcome

  • Delaware Vice Chancellor Lori W. Will sanctioned attorney Richard Rollo $2,500 under Court of Chancery Rule 11 for submitting a brief containing fabricated quotations, inaccurate citations, and mischaracterized legal authorities.
  • Rollo was required to pay the penalty personally within 14 days and file an affidavit confirming payment.
  • Richards, Layton & Finger was not sanctioned because the court found that the firm had established substantial AI safeguards and mandatory ethics training before the filing was submitted.
  • The court reaffirmed that attorneys have a personal, nondelegable duty to verify the accuracy of legal authorities and representations submitted to a court.
  • The ruling clarified that using generative AI in legal practice is not inherently prohibited, but lawyers remain accountable for errors in AI-assisted work.

Why It Matters

  • The decision reinforces that artificial intelligence does not replace a lawyer’s professional judgment or responsibility for court filings.
  • Fabricated quotations and inaccurate legal citations can undermine judicial proceedings and damage confidence in the legal profession.
  • Law firms may face different consequences from individual attorneys when they can demonstrate that reasonable safeguards were established and followed at the institutional level.
  • Mandatory AI ethics training, approved-tool restrictions, and human verification procedures are becoming increasingly important for legal organizations.
  • The ruling provides guidance for attorneys and law firms as generative AI becomes more common in legal research, drafting, and litigation.

 

 

 

 

 

 

 

 

author avatar
Janice Thompson
Janice Thompson enjoys writing about business, constitutional legal matters and the rule of law.

Janice Thompson

Janice Thompson enjoys writing about business, constitutional legal matters and the rule of law.