Oregon Lawyer Fined $100K for AI-Generated Legal Brief

A U.S. District Court in Oregon imposed a nearly $100,000 fine on attorney Steve Brigandi after legal filings submitted to the court contained numerous fictitious citations produced by artificial intelligence. Magistrate Judge Mark D. Clarke issued the ruling in a dispute tied to Valley View Winery, describing the AI-generated material as a “notorious outlier in both degree and volume” of hallucinations.

The sanction arose from a five-year inheritance fight involving the winery. The underlying lawsuit was filed in 2021 by Joanne Couvrette against her brothers, Mark and Michael Wisnovsky, in a conflict over ownership of the family business. Court records cited in the ruling show that the problematic submissions were not isolated: multiple filings made between January and May 2025 included fabricated legal precedents.

According to the ruling, opposing counsel warned about the false citations, but the filings continued. The court then intervened, dismissing Couvrette’s case against her brothers and penalizing Brigandi, with the judge pointing to the lawyer’s responsibility for how AI was used in the preparation and submission of court documents. The court also said there was evidence indicating Couvrette was directly involved in drafting the briefs that contained the fictitious citations.

The decision lands amid wider scrutiny of AI-generated text in legal practice, where courts rely on accurate citations and verifiable precedent. The Oregon ruling underscores that, regardless of drafting tools, attorneys remain accountable for what is filed under their names and for ensuring that cited authorities exist and support the arguments presented. In this case, the court’s language emphasized the scale of the hallucinations, framing the episode as unusually severe.

Beyond the immediate parties, the incident reflects a broader pattern being documented across the legal system. An online database tracking AI misuse in legal settings now lists over 1,300 cases, highlighting how frequently courts and litigants are encountering filings that include AI-related errors. The Oregon matter adds to that record, while also showing how warnings from adversaries may not prevent escalation if inaccurate material continues to be submitted.

Key uncertainties remain limited to what the ruling itself addressed: while the court cited evidence of Couvrette’s involvement in drafting, the decision focused the sanction on Brigandi’s responsibility for the filings. The case closes with the dismissal of Couvrette’s claims against her brothers and a financial penalty that signals heightened judicial intolerance for unverifiable AI-generated citations in court submissions.