Superb Motors v. Deo: $1,500 AI Sanctions Despite Computer Virus Defence | Advocate Prakhar

Case at a Glance

Full Citation	Superb Motors Inc. et al. v. Deo et al., 23-CV-6188 (JMW) (E.D.N.Y. April 30, 2026)
Court	U.S. District Court, Eastern District of New York
Date	April 30, 2026
Judge	James M. Wicks, U.S. Magistrate Judge
Category	AI Hallucination — Pro Se Attorney; Computer Virus Defence; $1,500 Sanction
Jurisdiction	USA — Federal (E.D.N.Y.)
AI Tool	Generative AI (not specified; Thomasson acknowledged use)
Outcome	$1,500 monetary sanction payable to court registry by May 15, 2026; no bar referral

Background

Superb Motors Inc. is a complex commercial litigation case in the Eastern District of New York involving automotive dealerships, involving extensive discovery disputes. Defendant Harry R. Thomasson, Esq. appeared pro se. In a brief filed in opposition to the plaintiffs’ motion to vacate a preliminary injunction, Thomasson included citations to fictitious cases — AI-generated hallucinations that had no basis in reality. The court identified the citations during review and issued a show-cause order. Thomasson’s response attributed the fabrications to viruses or malware that had “infiltrated” his computer, providing Staples receipts showing antivirus programs had been installed.

The AI Issue

Thomasson acknowledged using generative AI in drafting but focused his response on the computer virus theory rather than accepting direct responsibility for failure to verify citations. Judge Wicks found this approach unpersuasive. While the court acknowledged that Thomasson did not attempt to deny the sanctionable conduct once confronted, the virus defence failed to demonstrate that Thomasson had verified the citations before filing — which he had not. As an attorney, Thomasson was held to the full professional standard, not the more lenient standard sometimes applied to lay pro se litigants.

What the Court Decided

• AI-generated hallucinations submitted as valid caselaw constitute subjective bad faith — satisfying the higher standard required for sua sponte Rule 11 sanctions. [bad faith standard]
• A computer virus or malware infiltration does not negate an attorney’s responsibility to confirm work product before filing. [personal verification duty]
• Thomasson, as an attorney, is held to the full professional standard even when appearing pro se — the more lenient standard for lay self-represented parties does not apply. [attorney-as-pro-se standard]
• Monetary sanctions in AI cases have ranged from $1,000 to $5,000; a $1,500 sanction is appropriate given acknowledgement of severity and representation of future vigilance. [proportional sanction]
• No bar referral — Thomasson’s acknowledgement of severity and representation of future vigilance were mitigating factors. [no referral threshold]
• Sanction must be paid to the court registry by May 15, 2026. [deterrence payment]

“Courts in this circuit have repeatedly found that presenting AI-generated hallucinations as valid caselaw constitutes subjective bad faith.”

— Judge James M. Wicks, E.D.N.Y., April 30, 2026

The India Angle

Indian Law Equivalent

Under Indian law, the concept of “deliberate” or “wilful” misconduct is central to contempt under the Contempt of Courts Act 1971 — but unlike the US bad faith standard, Indian contempt law can be civil (failure to comply with orders) or criminal (scandalising the court). An advocate who files AI-hallucinated citations could face criminal contempt for knowingly presenting false legal authority, even if the initial error was AI-generated, particularly if discovered and not corrected.

Bar Council Rules

Rule 11 of the BCI Rules requires advocates to refrain from any conduct that is dishonest and incompatible with the dignity of the legal profession. The computer virus defence would likely fare no better before the BCI Disciplinary Committee than it did before Judge Wicks — the question is whether the advocate verified citations before filing, not what caused the original error.

Practical Advice for Indian Advocates

• Do not attempt to shift responsibility for hallucinated citations to technical failures — courts treat the verification duty as personal and non-delegable regardless of the root cause.
• A $1,500 equivalent in Indian currency would be meaningful for individual practitioners; more significant is the reputational cost of a published sanctions order, which Indian legal databases will increasingly index.
• When appearing pro se in any capacity — as advocates sometimes do in personal legal matters — apply the same verification standards as when representing clients; attorney status does not reduce obligations and cannot be disclaimed by claiming self-representation.

Quick Takeaways

• Computer virus defences for AI hallucinations are rejected — the duty to verify citations is personal and non-delegable.
• Attorneys appearing pro se are held to the full professional standard, not the lenient lay-litigant standard.
• The $1,000–$5,000 AI sanction range in the Second Circuit provides a predictable calibration framework.

---

Deep Dive: The $1,000–$5,000 AI Sanction Range and How Courts Calibrate It

Superb Motors v. Deo provides one of the most comprehensive judicial surveys of the AI sanction spectrum available in 2026. Judge Wicks synthesized precedent from across the Second Circuit and beyond to establish a benchmark: monetary sanctions for AI hallucinations range from $1,000 (lenient: first offence, prompt admission, cooperation) to $5,000 (Mata v. Avianca: repeated failure, multiple fabrications, inadequate response). The $1,500 figure in Superb Motors falls toward the lenient end because Thomasson acknowledged the severity of the issue and represented future vigilance, but above the floor because he initially blamed a computer virus rather than accepting direct personal responsibility.

This calibration framework is significant because it gives practitioners, clients, and insurers a predictable range within which AI misconduct sanctions will fall absent aggravating factors. More severe consequences — bar referral, compensation forfeiture, attorney removal, or appeal dismissal — apply when the AI misconduct is combined with additional factors: prior offences, failure to cooperate with the court’s investigation, deliberate intent to mislead, or harm to an indigent or vulnerable client. The basic monetary sanction tier is designed to be large enough to hurt but not so large as to be disproportionate to the typical solo practitioner’s billing rates.

The computer virus defence is worth examining as a category of response. Thomasson was not the first attorney to suggest that technical problems — corrupted files, software failures, data transfer errors — contributed to citation problems in AI-assisted documents. Courts have uniformly rejected such defences on the grounds that the attorney’s nondelegable verification duty exists regardless of how the error originated. The attorney’s job is not to produce an error-free first draft but to verify before filing — and that verification step is wholly within the attorney’s control irrespective of what happens upstream in the drafting process.

For Indian legal malpractice practitioners, the Superb Motors sanction range is a useful reference point as India develops its own AI liability framework. Professional indemnity insurers in India are beginning to exclude from coverage liabilities arising from unverified AI use — placing the $1,500 equivalent of a sanction directly on the individual advocate. As Indian courts develop their own AI sanction jurisprudence, the calibrated approach — proportional to harm, acknowledgement of responsibility, and deterrence — offers a template that Indian courts and bar councils can adapt within the framework of the BCI Rules and the Advocates Act.