Fuller v. Hyde School: Maine Court Orders CLE and Firm Policy Changes for AI Hallucinations | Advocate Prakhar

Case at a Glance

Full Citation	Fuller v. Hyde School et al., 2:25-cv-00354-SDN (D. Me. May 5, 2026)
Court	U.S. District Court, District of Maine
Date	May 5, 2026
Judge	Stacey D. Neumann, U.S. District Judge
Category	AI Hallucination — Education-Focused Non-Monetary Sanctions; CLE Order
Jurisdiction	USA — Federal (Maine)
AI Tools	Claude (Anthropic) or ChatGPT (OpenAI) — both named
Outcome	Non-monetary: CLE attendance, firm AI policy creation, client notification, brief stricken, compliance certification within 45 days

Background

Jessica Fuller brought a child exploitation claim against her former school, the Hyde School, and institutional defendants. After defendants moved to dismiss on November 7, 2025, Guagenty filed an opposition on November 25 using AI-generated content that included inaccurate, misleading, and potentially fabricated citations. When defendants flagged the errors, Guagenty filed a “Notice of Errata” — which itself introduced additional inaccuracies. The court reviewed the citations independently and found cases cited for unsupported propositions, language that could not be found in the cited sources, and inaccurate pincites. A show-cause order issued in April 2026 elicited Guagenty’s admission that she or a team member had used either Claude or ChatGPT.

The AI Issue

Guagenty’s response acknowledged for the first time (the court noted the delayed candour as an aggravating factor) that AI had been used in drafting. She described the process as an “admittedly poorly overseen” multi-attorney drafting exercise. She had not conducted a line-by-line citation check. When first confronted by defendants’ objections, she did not disclose AI use — she waited until the court’s show-cause order. She then apologized, accepted responsibility, retained new local counsel and additional firm oversight, implemented new AI policies, registered for an AI-specific CLE on April 29, 2026, and offered to file an amended opposition.

What the Court Decided

• The Notice of Errata that introduced further inaccuracies aggravated rather than mitigated the original violation. [errata-as-aggravation]
• Delayed candour — disclosing AI use only after a show-cause order — impairs the court’s ability to assess the source of errors and implicates the duty of candour. [timely disclosure obligation]
• AI is not inherently improper; it is the failure to verify AI output that triggers Rule 11 liability. [technology-neutral duty]
• Guagenty was admonished publicly; the admonition extends to the broader bar as a warning about verification obligations. [bar-wide admonition]
• Sanctions ordered: (1) submit proof of AI CLE attendance; (2) create and explain firm AI procedures to the court; (3) provide this order to the client; (4) file a compliance certification within 45 days. [educational and systemic sanctions]
• The defective opposition brief stricken from the docket; plaintiff ordered to file an amended response. [brief striking and refiling]

“While technology may evolve, the attorney’s nondelegable duty to exercise independent professional judgment and verify all representations made to the tribunal remains a cornerstone of our legal system — one that cannot be sacrificed to the convenience of automation.”

— Judge Stacey D. Neumann, D. Maine, May 5, 2026

The India Angle

Indian Law Equivalent

India’s Continuing Legal Education framework under the Bar Council of India rules is less developed than the US CLE system. However, Section 49(1)(a) of the Advocates Act 1961 empowers the BCI to prescribe standards of professional conduct. A BCI directive requiring advocates who have committed AI hallucination errors to attend AI-specific training before filing again would be within its statutory authority and consistent with the Maine court’s education-focused approach.

Bar Council Rules

Rule 23 of the BCI Rules requires advocates to keep abreast of changes in the law. The Fuller court’s CLE order operationalizes this duty — requiring formal education on AI limitations and responsible use as a remedial sanction rather than leaving the learning to happen informally. Indian state bar councils should consider making AI literacy CLE mandatory for advocates practicing in high-volume digital document courts.

Practical Advice for Indian Advocates

• Develop and document firm-level AI use policies before a court asks you to — the Fuller court made policy creation a sanction condition; having a policy in place before any incident is both a compliance advantage and a professional responsibility marker.
• Disclose AI use to the court proactively when it is relevant to citation accuracy — delayed disclosure, as in Fuller, is treated as a breach of candour obligations and may convert a correctable error into a disciplinary matter.
• Treat a Notice of Errata or correction filing with as much care as the original — introducing new errors while attempting to correct old ones compounds the violation and undermines judicial confidence in the attorney’s competence.

Quick Takeaways

• A Notice of Errata that introduces new errors compounds the original AI hallucination violation.
• Proactive disclosure of AI use is a candour obligation — waiting for a show-cause order to disclose is treated as an aggravating factor.
• Education-focused sanctions (CLE, firm policies, compliance certification) are an emerging model for first-time AI hallucination offenders.

---

Deep Dive: Education as Sanction — The Emerging Regulatory Model for AI Misconduct in Law

Fuller v. Hyde School belongs to a growing category of AI sanctions rulings that prioritize education and systemic reform over pure punishment. This approach reflects a judicial recognition that the hallucination crisis is, at least in part, an information problem: many attorneys do not fully understand how large language models work, why they hallucinate, and what verification steps are required to make AI output safe to file in court. The Fuller court’s response — requiring attendance at an AI-specific CLE, creation of firm-level policies, and certification of compliance — treats the violation as remediable through learning rather than purely punishable through financial penalty.

This educational sanction model is modelled, at least implicitly, on how courts have historically addressed other technology-competence failures in the legal profession. When electronic filing was mandated, courts developed training programs and compliance schedules for attorneys who filed defective electronic documents. When digital evidence became central to litigation, courts began requiring attorney certification of e-discovery competence. The AI hallucination CLE requirement follows the same logic: a new technology has created new risks, and attorneys must be formally educated in managing those risks as a condition of continued practice.

The distinction between Claude (Anthropic) and ChatGPT (OpenAI) named in the Fuller court order is noteworthy. These are the two most widely used general-purpose AI assistants among legal professionals who lack access to purpose-built legal AI tools. Their hallucination rates and citation behaviours differ — studies comparing legal AI tools consistently find that general-purpose models like these are more prone to citation fabrication than purpose-built legal research AI such as CoCounsel or Lexis+ AI. Guagenty’s case illustrates the risk of using a general-purpose chatbot — optimized for fluency and helpfulness rather than legal accuracy — to draft documents that will be submitted to a federal court.

For Indian law firms building AI governance frameworks, the Fuller order provides a template: train your team before an incident, not after. Specifically, firms should identify which AI tools their attorneys are using (general vs. legal-specific), assess the hallucination risks associated with each tool class, implement verification protocols that are tool-specific, and maintain training records that can be produced if a court ever inquires. The BCI Rules as currently written do not mandate this — but the global trajectory of AI regulation in the legal profession makes it clear that such requirements are coming, and proactive firms will be better positioned when they arrive.