Rushing v. Turner: Attorney’s EVE AI Hallucination and Candid Court Apology | Advocate Prakhar

Case at a Glance

Background

Rushing v. Turner is a personal injury case in Louisiana federal court in which attorney Ross LeBlanc represented the plaintiff. To support motions to compel additional financial data from two medical experts, LeBlanc filed briefs citing Riggio v. Ports Am. La., L.L.C., 337 So.3d 418 (La. App. 4 Cir. 2023), for the proposition that “evidence of the financial relationship between an expert witness and the party who engaged the expert is always admissible to show bias.” The substantive proposition is legally correct — but the precise quoted language does not appear anywhere in the Riggio decision. Opposing counsel caught the error and raised it in their opposition.

The AI Issue

LeBlanc disclosed in his apology letter that he had begun using EVE AI, a legal drafting tool marketed with a “Trust but Verify” feature that provides hyperlinks to cited quoted material. After initially checking EVE’s citations frequently and finding them accurate, LeBlanc grew confident in the tool and reduced his verification frequency. He believes the fabricated quotation resulted from either EVE’s AI output or a copy-paste error on his part. The apology letter was notable for its transparency — LeBlanc did not blame the software but accepted full professional responsibility, promised to review all recent filings for similar errors, and circulated the letter to colleagues as a teaching moment.

What the Court Decided

• LeBlanc voluntarily withdrew the two motions containing the inaccurate citation before any formal court ruling on sanctions. [curative withdrawal]
• The substantive legal proposition was correct; only the quotation marks were the problem. [de minimis vs. structural error]
• Opposing counsel requested striking of pleadings and an award of attorney fees and costs. [Rule 11 sanctions motion]
• LeBlanc accepted full responsibility and committed to reviewing all pending filings in that court for similar errors. [proactive remediation]
• The case highlighted that AI tools with built-in verification features are not foolproof — complacency develops when initial checks consistently succeed. [over-reliance doctrine]

“I never thought this could happen to me. I cannot be entirely sure this mistake involved the EVE AI software, or if this was my personal mistake in using cut/copy/paste. Either way, the mistake is ultimately mine and I understand that.”

— Attorney Ross LeBlanc, Apology Letter to Judge WilliamJorden, March 27, 2026

The India Angle

Indian Law Equivalent

Indian advocates filing expert witness challenge motions under Order XVIII Rule 17 CPC (expert examination) or Section 45 Indian Evidence Act must accurately cite precedents supporting financial bias challenges to medical experts. Any fabricated citation in such filings could be treated as contempt under the Contempt of Courts Act 1971 or professional misconduct under Section 35 of the Advocates Act 1961.

Bar Council Rules

Rule 33 of the Bar Council of India Rules mandates that advocates conduct their cases by fair and honourable means. Rule 11 prohibits acts of misconduct. The LeBlanc apology model — immediate withdrawal, transparent disclosure, proactive review, and internal firm circulation — represents best practice compliance that Indian advocates should adopt when discovering similar errors.

Practical Advice for Indian Advocates

• Never trust AI “verify” hyperlinks as a substitute for reading the actual judgment — always open the original source and confirm the exact quoted language appears in the text.
• Complacency is a specific risk: initial success with AI verification does not guarantee future accuracy; maintain consistent cross-checking on every filing.
• If a citation error is discovered post-filing in Indian courts, file a memo of corrections immediately and inform opposing counsel — proactive disclosure mitigates disciplinary consequences.

Quick Takeaways

• AI tools marketed with verification features can still hallucinate; human review remains mandatory.
• Voluntary withdrawal and candid disclosure significantly mitigate sanction risk.
• Circulating the error internally as a teaching moment is professional best practice.

Deep Dive: The EVE AI Case and the Psychology of AI Complacency in Law

Rushing v. Turner is unusual in the AI hallucination literature because of what it reveals about the psychology of AI adoption in legal practice. Attorney Ross LeBlanc was not a reckless early adopter who blindly trusted an unfamiliar tool. He was cautious, sceptical, and initially diligent. He conducted the legal research himself, restricted EVE to approved reference materials, and checked the AI’s citations frequently during the early weeks of use. The problem was that his early checks always confirmed accuracy — building a dangerous sense of confidence that ultimately led to reduced verification frequency.

This pattern — initial caution giving way to habitual trust — is documented in human factors research as “automation complacency.” Pilots who over-rely on autopilot systems, radiologists who trust AI screening tools without independent review, and now lawyers who stop verifying AI citations after a period of apparent reliability are all exhibiting the same cognitive pattern. The legal profession’s adoption of AI tools is accelerating this risk because legal drafting is cognitively demanding, and any tool that reduces that burden creates strong incentives to reduce oversight.

EVE AI’s “Trust but Verify” marketing language is itself instructive. The phrase acknowledges that AI output requires verification — but the product design, which provides a single hyperlink per citation, may create the false impression that verification is already built in. A lawyer who clicks a link and sees a case that genuinely exists may not then read the full text to confirm the specific quoted language appears. LeBlanc’s case illustrates exactly this gap: the case was real, the legal proposition was substantively correct, but the quotation marks around specific language invented by the AI were the source of the problem.

For Indian advocates, the lesson extends to every AI-assisted legal research tool available in the Indian market — whether domestic platforms or internationally licensed tools. The Bar Council of India has not yet issued specific AI use guidelines, but the principles are clear under existing Rules of Professional Conduct: candour toward the tribunal is non-negotiable, and the duty to verify cannot be delegated to software. LeBlanc’s voluntary apology and proactive internal review represent the gold standard response — a model that Indian legal firms should build into their AI governance policies now, before a similar incident reaches an Indian court.