How much was the sanction in Couvrette v. Wisnovsky?

$110,204, the largest U.S. AI-hallucination sanction on record as of 2026, split between pro hac vice counsel and local counsel as monetary sanctions plus the opposing side's attorney fees.

← AI sanctions tracker

Free tools · Sanctions tracker · Oregon

Couvrette v. Wisnovsky

Q: What did the lawyer do wrong?

Three summary-judgment briefs filed over about five months contained 15 fabricated case citations and 8 fabricated quotations. After opposing counsel flagged the errors, the lawyer refiled without verifying the arguments that remained.

$110,204

U.S. District Court, District of Oregon·2026·Largest U.S. AI-hallucination sanction on record

Court

U.S. District Court, District of Oregon

Year

2026

Penalty

$110,204, split between pro hac vice counsel and local counsel as monetary sanctions plus the opposing side's attorney fees

What happened

Three summary-judgment briefs filed over roughly five months contained 15 fabricated case citations and 8 fabricated quotations. After opposing counsel flagged the errors, the lawyer refiled without verifying the arguments that remained.

Primary source

ABA Journal report →

The duty behind the penalty

The court did not punish the tool. It punished the missing review.

What drew the sanction was not that an AI tool produced bad text. It was that briefs went to the court carrying citations no one had confirmed, and then went back a second time still unverified. The obligation to check what you file is the lawyer's, and it does not transfer to whatever drafted the words. The expensive part of this case is the part a firm can actually control: a record of what was checked, by whom, before it went out.

See how Candor builds that record Back to the tracker

Common questions

How much was the sanction?

$110,204, the largest U.S. AI-hallucination sanction on record as of 2026, split between the pro hac vice counsel and local counsel plus the opposing side's attorney fees.

What was the underlying failure?

Fifteen fabricated citations and eight fabricated quotations reached the court across three briefs, and a refiling repeated the pattern rather than verifying what remained. The consistent thread across these cases is an unreviewed citation, not the use of AI itself.

This is a reference summary of a publicly reported decision, compiled from the source linked above. It is not legal advice and is not exhaustive. Confirm the details against the primary source. Candor is software, not a law firm.