Burches v. Equifax: $500 Sanction for Repeat AI Hallucinations After Express Court Warning

Background: A Pro Se Plaintiff, a Credit Dispute, and a Pattern of Fake Citations

Taiujuan Burches filed suit in the Eastern District of Louisiana against multiple credit reporting and financial services defendants — Equifax, Trans Union, and Comenity Bank — in what began as a routine consumer credit dispute. What distinguished this case was not the underlying claims but what emerged from Burches’s filings: a disturbing pattern of fabricated, inaccurate, and unverifiable legal citations appearing across multiple briefs.

The problem first surfaced when Trans Union filed an ex parte motion for leave to file a surreply, specifically noting that “a closer examination of Plaintiff’s Motion reveals Plaintiff’s reliance on fabricated and/or non-existent legal authorities.” The court granted the motion. Before oral argument, Burches filed a Notice of Clarification admitting that the citations in his reply brief were fabricated and apologizing for the error.

At the oral argument, Judge Morgan addressed the fabricated citations directly. Citing Burches’s pro se status, the court declined to impose sanctions at that time — but issued an express warning: any future filings with fabricated citations would result in personal sanctions. The court also ordered Burches to review all his prior filings and report back by November 24, 2025, identifying any other problematic citations. Burches complied in part, sending an email on November 23, 2025, identifying three additional incorrect citations across his prior filings.

The AI Issue: Post-Warning Recidivism

Despite the warning, the self-audit, and the apology, Burches did not stop. He filed an Opposition to Comenity Bank’s Motion to Compel Arbitration containing additional incorrect citations and fabrications. The court was specifically unable to verify the quoted language he attributed to the Fifth Circuit in Will-Drill Resources, Inc. v. Samson Resources Co., or the purported holding of Hays v. HCA Holdings, Inc.

On March 2, 2026, the court issued a show-cause order requiring Burches to appear and explain why sanctions should not be imposed. Rather than stopping, Burches filed yet another memorandum — this time opposing Trans Union’s Motion to Stay Discovery — containing a citation the court could not verify. This occurred after the show-cause order was already on the docket.

At the May 11, 2026 hearing, Burches admitted that he had signed and filed multiple briefs without verifying the accuracy of the citations contained in them. The court applied an objective standard of reasonableness under Rule 11 and found the conduct fell below that standard. The repeated submission of false or unverifiable authority — especially after an express warning, a self-audit, and a show-cause order — undermined judicial confidence and imposed unnecessary burden on the court and opposing parties.

Holdings and Sanctions

Judge Morgan issued four key rulings:

1. Rule 11(b)(2) violated. Burches violated his duty to certify that the legal contentions in his filings were warranted by existing law or a non-frivolous argument for modifying existing law. The objective reasonableness standard applies equally to pro se litigants.
2. Post-warning conduct is independently sanctionable. The court emphasized that the critical factor was not just the initial fabrications but the continuation after an express judicial warning — signaling that the court views post-notice recidivism as a serious aggravating factor.
3. $500 sanction imposed. The monetary sanction was payable by June 10, 2026, to the Clerk of Court.
4. Future violations may result in dismissal. The order explicitly warned that further fabricated authorities, miscited holdings, or false quotations “may result in sanctions, including dismissal of this action.”

“The repeated submission of false or unverifiable legal authority, particularly after an express warning from the Court, falls below the objective standard of reasonableness required by Rule 11. Such conduct undermines the Court’s confidence in Mr. Burches’s filings, imposes unnecessary burdens on the Court and opposing parties, and interferes with the orderly administration of justice.”

— Judge Susie Morgan, E.D. Louisiana, May 11, 2026

India Angle: What Indian Lawyers and Litigants Must Know

The Burches case holds a particular lesson that resonates deeply in the Indian legal context: the danger is not just in using AI to generate citations — it is in continuing to do so after being warned. Indian courts may encounter similar patterns as AI tools become more accessible to pro se litigants filing under CPC or consumer forums under the Consumer Protection Act 2019.

Relevant Indian Law

• Contempt of Courts Act 1971, Section 2(b) & (c): Persistent filing of false or fabricated citations, especially after a court warning, constitutes criminal contempt — an act that “scandalises” the court or “substantially interferes with the course of justice.” A single fabrication might attract civil contempt; a pattern of post-warning fabrication could cross into criminal contempt territory.
• BCI Rules of Professional Conduct, Rule 9: An advocate must not act “in a manner tending to mislead a court.” Submitting fabricated citations is a textbook Rule 9 violation. For advocates (unlike Burches who was pro se), the BCI can additionally initiate professional misconduct proceedings under Advocates Act 1961 Section 35.
• Code of Civil Procedure 1908, Order VI Rule 16: Courts have power to strike pleadings containing scandalous or vexatious matter. While this rule targets pleadings, courts invoking inherent powers under Section 151 CPC have wider latitude to impose costs on parties who abuse the process — including through fabricated citations in applications and arguments.

Three Practical Tips for Indian Practitioners

1. Verify every citation before filing, not after. Burches’s downfall was signing briefs he had not verified. Under Bar Council Rules and judicial ethics, every advocate is personally responsible for citations submitted to the court. If AI suggests a case, independently locate it on SCC Online, Manupatra, or Indian Kanoon before citing it — there is no AI-generated citation exemption.
2. If you discover an error, correct it immediately and formally. Burches admitted the fabrications before oral argument, which initially saved him from sanctions. Indian practitioners facing a similar situation should file a correction memo the moment they discover an AI error, citing RPC Rule 22 (duty to inform the court of relevant adverse authority) by analogy. Prompt disclosure is a strong mitigating factor.
3. Treat a judicial warning as a bright line, not a caution. The moment a court tells you that fabricated citations will result in sanctions, treat it as an absolute prohibition. In India, where judicial displeasure can translate to contempt proceedings with swifter consequence than a U.S. Rule 11 process, a post-warning violation is especially dangerous for an advocate’s practice and standing.

Quick Takeaways

• Pro se status does not shield a litigant from Rule 11 sanctions for AI hallucinations — an objective reasonableness standard applies to everyone.
• Continuing to file fabricated citations after an express judicial warning and a court-ordered self-audit is a serious aggravating factor that justifies monetary sanctions.
• Filing fabricated citations after a show-cause order is already pending may accelerate the court toward dismissal as the ultimate sanction.
• The court’s “verify before signing” duty is unambiguous: an attorney or pro se litigant must personally confirm citations — AI is not an excuse.
• In India: repeat AI citation errors + court warning = potential contempt, not just a professional reprimand.

Deep Dive: The Escalating Sanctions Ladder in AI Hallucination Cases

Burches sits at a critical point on what courts are increasingly treating as a sanctions escalation ladder: (1) first-time AI hallucination + no prior warning → warning only; (2) first-time AI hallucination + acknowledgment and apology → warning + self-audit order; (3) repeat AI hallucination after warning → monetary sanction; (4) repeat AI hallucination after show-cause order → dismissal or heavier sanction.

This escalation pattern is visible across recent AI hallucination cases. In Allen v. Hunt (N.D. Illinois 2026), a first-offense citation error resulted only in a warning. In Fogarty v. Fogarty, a $2,500 sanction was imposed for a second-offense pattern. In Couvrette v. Wisnovsky (Oregon 2026), 15 fabricated cases across three briefs resulted in terminating sanctions — the harshest outcome yet recorded. Burches fits squarely in the middle tier: $500 monetary sanction, but with a clear warning that dismissal is next.

The lesson for practitioners — Indian and American alike — is that courts are not merely punishing AI use; they are punishing the failure to verify. An advocate who uses AI but personally verifies every citation has not violated Rule 11. An advocate who signs briefs without verification has violated it, regardless of whether a human or a machine wrote the first draft. Burches makes that duty explicit: “the duties imposed by Rule 11 require that attorneys read, and thereby confirm the existence and validity of, the legal authorities on which they rely.”

For Indian advocates, the verification duty maps onto BCI Rule 14 (duty to maintain accuracy of legal submissions) and the broader framework of the Advocates Act 1961. As Indian courts begin to encounter AI-assisted filings — especially from pro se litigants using tools like ChatGPT or Gemini to draft consumer forum complaints or writ petitions — the Burches pattern is likely to recur. Indian courts would be well within their contempt jurisdiction to impose progressively heavier consequences on litigants who continue to submit fabricated citations after a judicial warning.