Miller v. Regions Bank (N.D. Ala. May 21, 2026)

Why This Case Is Important

While courts frequently extend leniency to technology-related errors when counsel acts with candor, they will aggressively punish intentional digital spoliation and bad-faith concealment.

Overview Text

In Miller v. Regions Bank, a routine employment matter transformed into a severe ethical and E-Discovery crisis. The issue began when Plaintiff’s attorney submitted an opposition brief fighting a summary judgment motion. The brief featured four highly substantive legal quotations attributed to legitimate appellate precedents. However, an independent review by the court exposed a glaring issue: while the case names were accurate, the text inside the quotes was entirely fabricated—a direct byproduct of generative artificial intelligence "hallucinations."The court issued a show-cause orders under Rule 11, demanding plaintiff’s attorney explain his legal research methods. Instead of admitting to an AI blunder, the attorney engaged in what the court characterized as a shifting array of distortions and evasions. Harp initially denied usingAI to draft any part of the brief, asserting that the tool had not written "a single word." The court subsequently ordered plaintiff’s attorney to produce his comprehensive ChatGPT history, but he informed the court that the account had been deleted and the files were unavailable. In fact, after receiving the preservation order, Harp intentionally deleted his ChatGPT account and sought an immediate refund to ensure the logs were permanently destroyed.

Ruling Summary

• AI Spoliation Establishes Adverse Inference The court applied traditional spoliation principles and its inherent authority to penalize the intentional purge of electronic accounts. The court found that deleting active logs directly after receiving a judicial directive "supports a strong inference that Attorney Harp acted in bad faith." Consequently, the court levied an adverse inference, stating that the missing account histories "would have shown that ChatGPT generated the fabricated quotations and related legal misrepresentations contained in the Response Brief.”  
• ‍Sanctions Tied to Deception, Not Modern Technology The court took care to separate technological adaptation from ethical misconduct. Judge Mooty explicitly clarified that the bench was "not sanctioning [plaintiff’s attorney] merely for using generative AI. Attorneys may use such tools responsibly." Rather, the severe penalties stemmed squarely from the "submission of false legal authority, lack of reasonable inquiry, lack of candor to the court, and destruction of evidence."
• ‍Sweeping, Career-Altering Penalties Levied Stating that "competent and ethical lawyers own" their errors, the court noted the cover-up was, in this instance, worse than the crime. “When lawyers are caught submitting AI-generated misrepresentations to the court, they have two options: they can either admit to their mistakes and show contrition, or they can attempt to cover up their mistakes and demonstrate a weakness of character unsuited to the legal profession.” The court imposed "career-altering sanctions" including a public reprimand, disqualification from the Miller action, a six-month suspension from local practice, mandatory notification to all current clients and active judges, and a formal referral to state licensing bodies.

Expert Analysis

Erin Corken, Esq., CEDS, FIP, AIGP, CIPP/US, CIPP/E, CIPM, Senior Solutions Engineer, Exterro

Two things did not go well here. First, the attorney used AI in an unethical manner because he did not check the output as required. Under the framework of ABA Formal Opinion 512, blindly relying on generative AI violates Model Rule 5.1 and Model Rule 5.3 regarding the duty to supervise nonlawyer assistance. These rules required him to review the AI's work just as he would a human assistant's, rather than relying on it uncritically. Second, he tried to cover up the mistake by deleting his account logs and misleading the court, violating Rule 3.3 (Candor toward the Tribunal).

It’s so important that legal professionals stop now in their tracks and learn what their ethical obligations are regarding using AI and follow them. This is not as easy as it sounds because the landscape is changing rapidly. Courts are issuing their own local rules, so each judge might require something different regarding disclosure… if they permit using AI at all. Legal teams must check what the rules are for every matter, every time, and follow them.

We can sympathize with Harp though. I think if people don’t understand whether using AI is permitted—or any specific restrictions or disclosure requirements that may exist—they get caught off guard when questioned about it. Out of fear, they can panic and deny it. The line that stands out to me the most from the decision is: “Lawyers make errors. Competent and ethical lawyers own them.”

Tip Text

AI prompt streams can be discoverable ESI. Cover-ups in the digital age are readily detectable and treated with zero tolerance. For an updated overview of discovery obligations, corporate legal teams should consult Exterro’s Guide to the FRCP.