Essandoh v. Capital One Bank NA
U.S. District Court, Eastern District of Texas, Sherman Division · E.D. Tex. · Texas bar guidance
Conduct
A pro se plaintiff's part of a joint discovery report contained inaccuracies he admitted at a hearing came from relying on generative AI.
Consequence
No sanction. Here the magistrate judge admonished him and entered a case order requiring an AI-use certificate on any AI-assisted filing.
Lesson
Here an admission of AI-caused errors at a discovery hearing led to a same-day AI certification order entered on the case docket rather than a sanction.
Verified September 28, 2026
- Citation
- Essandoh v. Capital One Bank NA, No. 4:25-cv-00746-ALM-BD, Order on Disclosure and Certification Requirements for Use of Generative Artificial Intelligence (E.D. Tex. Apr. 16, 2026) (Davis, M.J.), ECF No. 27
- Decided
- April 16, 2026
Summary
Laud Essandoh, proceeding pro se, sued Capital One Bank NA; the case was transferred from the Northern District of Texas to the Sherman Division of the Eastern District of Texas and referred to Magistrate Judge Bill Davis. On April 16, 2026, Judge Davis held an in-person discovery hearing on the parties' joint report. According to the minute entry and the order entered the same day, the court questioned Essandoh about inaccuracies in his part of the joint report, and "Essandoh admitted that the inaccuracies resulted from reliance on generative artificial intelligence ('AI')." The court admonished him about the risks of that reliance and the sanctions that can result. The minute entry records that the court "did not impose sanctions today but will issue an order regarding AI use that provides further guidance and admonishments." The order does not describe the inaccuracies as fabricated citations; it refers only to "inaccuracies" in the joint report.
- AI tool:
- Generative AI (the pro se plaintiff admitted at an April 16, 2026 hearing that inaccuracies in his part of a joint report resulted from reliance on generative AI; tool not identified)
What did the court warn about?
The same day, Judge Davis entered an Order on Disclosure and Certification Requirements for Use of Generative Artificial Intelligence on the case docket. Noting that the court's local rules (AT-3(m) and CV-11(g)) already address AI, the order supplements them in three ways. Any document filed that was prepared with AI assistance must include a "Certificate of Generative Artificial Intelligence Usage" that discloses the AI use, identifies each tool and how it was used, and states that the filer has confirmed the accuracy of every part of the filing prepared with AI; a filing without the certificate is assumed to have been prepared without AI. The attorney or pro se litigant who signs an AI-assisted filing is responsible for its content under Rule 11 and applicable professional-conduct rules. And if the court has good cause to believe an attorney or pro se litigant has relied on AI in violation of the order or any applicable rule, it "may impose appropriate sanctions, including striking a filing, dismissing a claim or defense, dismissing a case, and ordering monetary sanctions." No sanction was imposed.
Why does Essandoh v. Capital One Bank NA matter for law firms using AI?
Essandoh is a case-management response to AI misuse rather than a sanctions ruling. The problem arose in a joint discovery report filed after the court struck Essandoh’s motion to compel for noncompliance with the court’s rules, in the order the court later calls its “face-to-face order” (Dkt. 24). At the April 16, 2026 hearing, Essandoh admitted that the inaccuracies in his portion of the report came from reliance on generative AI. Judge Davis admonished him, imposed no sanction, and entered a written order the same day to supplement the hearing.
After reciting the hearing, the order quotes Mata v. Avianca describing the harms bogus citations cause opposing parties, courts, clients, and the justice system, and then adds three requirements to the Eastern District of Texas’s existing local rules: a certificate of generative AI usage on any AI-assisted filing, personal Rule 11 responsibility for the signer, and a menu of possible sanctions if the court has good cause to believe AI was relied on in violation of the order or the rules.
The certificate requirement is broader than a disclosure checkbox. It requires the filer to identify each tool, explain how it was used, and state that every AI-assisted part of the filing, including every factual, procedural, or legal assertion or citation, has been confirmed. The order also sets a default rule in its own words: “The court will assume that any filing that does not contain this certification was prepared without any use of AI.”
Implications for your firm
Operational steps a firm reading this case may wish to consider documenting. Strategic and rule-application calls belong to your firm's attorneys.
- Review joint reports and other non-brief filings for AI-generated inaccuracies. Here the problem surfaced in a party's section of a joint discovery report, not in a brief.
- Document AI use in filings before this court in the form the order specifies: disclosure, each tool and how it was used, and confirmation of the accuracy of every AI-assisted part.
- Consider the order's default rule: "The court will assume that any filing that does not contain this certification was prepared without any use of AI."
Sources
Primary sources
- The nature of the 'inaccuracies' in the joint report (Dkt. 25) is not described in the order or minute entry; whether they included citations to nonexistent authority was not verified.