Table of Contents
The Story That Should Have Every Managing Partner Paying Attention
A family inheritance dispute over an Oregon vineyard just produced what may be the largest sanction yet issued against lawyers for filing AI-generated “slop” in court — about $110,000 in sanctions and fees, a dismissed case, and a federal judge’s withering rebuke. [1] Damien Charlotin AI Hallucination Cases Database View source ↗
According to the court’s opinions, the case involved Valley View Winery, an 80-acre vineyard in southern Oregon [2] The New York Times A Family Feud at an Oregon Winery Turns to Vinegar Over A.I. Slop View source ↗ at the center of a five-year inheritance battle between four siblings. One sister, working with a California attorney representing her pro bono, submitted filings riddled with fabricated case citations — two in January 2025, seven more in April, and six more in May along with ten fabricated quotations, even after opposing counsel had already flagged the earlier fakes. [3] U.S. District Court, D. Or. Couvrette v. Wisnovsky, No. 1:21-cv-00157-CL, Opinion and Order (ECF No. 215) View source ↗ One of the fake citations pointed to an aggravated murder case unrelated to winery contracts or fee recovery. [2] The New York Times A Family Feud at an Oregon Winery Turns to Vinegar Over A.I. Slop View source ↗
The Court's Response
Magistrate Judge Mark D. Clarke called it “a notorious outlier in both degree and volume” of AI hallucinations. [3] U.S. District Court, D. Or. Couvrette v. Wisnovsky, No. 1:21-cv-00157-CL, Opinion and Order (ECF No. 215) View source ↗ He dismissed the sister’s case entirely, ordered her attorney to pay almost $100,000 in sanctions and fees, and ordered her local Oregon counsel to pay another $14,000 for failing to participate meaningfully in the case as local counsel. [3] U.S. District Court, D. Or. Couvrette v. Wisnovsky, No. 1:21-cv-00157-CL, Opinion and Order (ECF No. 215) View source ↗ [4] U.S. District Court, D. Or. Couvrette v. Wisnovsky, No. 1:21-cv-00157-CL, Opinion and Order (ECF No. 225) View source ↗
This isn’t an isolated incident. And that’s the part every law firm needs to understand.
The Scale of the Problem
The Valley View case is recorded in an online database maintained by Damien Charlotin, a legal researcher at HEC Paris who tracks court decisions in cases where generative AI produced hallucinated content. The database has been featured in news media and in several court decisions. [1] Damien Charlotin AI Hallucination Cases Database View source ↗
| Timeframe | Documented Cases of AI Hallucinations in Court Filings |
|---|---|
| Early November 2025 | ~500 |
| Mid-April 2026 | 1,300+ |
That’s more than two and a half times as many documented incidents in roughly five months. [1] Damien Charlotin AI Hallucination Cases Database View source ↗ The database catalogs fabricated citations, false quotes from real cases, misrepresented holdings, outdated or reversed authorities, and other AI-generated inaccuracies. It tracks whether the offender was a lawyer, a law firm, or a self-represented litigant — and what sanctions followed, from warnings and voided briefs all the way up to five- and six-figure fines. [1] Damien Charlotin AI Hallucination Cases Database View source ↗
Judicial Patience Is Wearing Thin
Historically, the consequences for AI misuse in filings have been modest: a warning, a voided brief, mandatory continuing legal education. [1] Damien Charlotin AI Hallucination Cases Database View source ↗ The Valley View sanction signals that the era of gentle corrections is ending.
What Actually Went Wrong in the Oregon Case
The failure points in the Valley View case are instructive because they map directly onto risks any firm using generative AI could face:
1. The attorney didn’t verify what he was signing.
The court found “persuasive” evidence that the client herself had likely written the offending briefs, which her attorney then stamped and filed. [3] U.S. District Court, D. Or. Couvrette v. Wisnovsky, No. 1:21-cv-00157-CL, Opinion and Order (ECF No. 215) View source ↗
2. The errors compounded because no one was checking.
Two fake citations in January became seven in April, then six more fake cases and ten fabricated quotations in May — after opposing counsel had already flagged the problem. [3] U.S. District Court, D. Or. Couvrette v. Wisnovsky, No. 1:21-cv-00157-CL, Opinion and Order (ECF No. 215) View source ↗ There was no review process catching the errors before filing.
3. Medical and personal circumstances didn’t save anyone.
The primary attorney’s doctor later submitted a declaration explaining that severe kidney disease had “significantly impaired” his cognitive function during the relevant period. The judge declined to give it full weight and held him accountable anyway. [3] U.S. District Court, D. Or. Couvrette v. Wisnovsky, No. 1:21-cv-00157-CL, Opinion and Order (ECF No. 215) View source ↗
Why This Matters to Professional Services Firms
Law firms are not the only organizations now using generative AI to draft documents, summarize materials, or accelerate research. Accounting firms, consultancies, and other professional services organizations face the same fundamental risk: AI tools produce confident-sounding output that can be completely fabricated, and the person who signs the final work product owns the consequences.
What the Valley View Case Establishes
- Courts are now willing to impose six-figure sanctions for AI hallucinations.
- “I didn’t write it” is not a defense.
- Health, time pressure, or reliance on a client’s research does not excuse the failure to verify.
- The supervising or local attorney is on the hook even if they weren’t the one using the AI.
A Practical Framework for AI Use in Your Firm
Based on the patterns documented across the 1,300+ cases in the hallucination database — and what we’ve learned from working with law firms on technology governance — here’s what we recommend professional services organizations put in place:
1. Establish a Written AI Use Policy
Every firm should have a policy that addresses, at minimum:
- Which AI tools are approved for firm use
- What types of work product can and cannot be generated with AI assistance
- The firm’s position on client data and confidentiality in AI tools
- Mandatory verification requirements for any AI-generated content used in client deliverables
2. Require Citation Verification
Any document containing legal citations, statutory references, or factual claims generated with AI assistance should pass through a mandatory verification step. This is non-negotiable. The technology exists — including automated citation-checking tools built specifically to detect hallucinated legal references — but policy and workflow matter more than any single tool.
3. Implement a “Two Sets of Eyes” Rule
The Valley View co-counsel’s fate is a cautionary tale. Anyone whose name appears on a filing, deliverable, or client communication is responsible for its accuracy — regardless of who drafted it or what tools they used. Build review checkpoints into the workflow before anything goes out the door.
4. Train Staff on AI Limitations
The most dangerous AI users are the ones who don’t understand how these tools fail. Generative AI doesn’t know what it doesn’t know. It can fabricate citations, misquote real cases, and invent plausible-sounding facts with complete confidence. Your team needs to understand this at a gut level, not just intellectually.
5. Audit and Log AI Use
In a regulated or high-stakes environment, you need to know what was used, when, and by whom. This matters for quality control, for client communication, and — if something goes wrong — for your ability to understand what happened and respond.
6. Review Your Insurance Coverage
AI-related errors are a new and evolving risk category. Talk to your carriers about how your current policies respond to AI-generated errors in work product, and whether your professional liability coverage addresses this category of risk specifically.
Where DP3 Fits
We work with law firms and professional services organizations in New York and Chicago on exactly these kinds of governance, policy, and technical control questions. If you’re thinking about how to roll out AI tools responsibly — or if your team is already using them and you want to make sure you have appropriate guardrails in place — we’d be glad to have a conversation.
The firms that will come out ahead aren’t the ones avoiding AI. They’re the ones using it thoughtfully, with the right policies, the right workflows, and the right verification built in.
The Valley View case is a preview of what happens when you skip those steps.
Contact us to discuss AI governance for your firm
References and Standards
- [1] Damien Charlotin, AI Hallucination Cases Database, updated September 25, 2026. Accessed September 25, 2026. https://www.damiencharlotin.com/hallucinations/ Case counts for November 2025 and April 2026 are from Internet Archive captures of this page dated November 3, 2025 (507 cases) and April 16, 2026 (1,317 cases).
- [2] Evan Gorelick and Anna Griffin, A Family Feud at an Oregon Winery Turns to Vinegar Over A.I. Slop, The New York Times, April 17, 2026. Accessed September 28, 2026. https://www.nytimes.com/2026/04/17/us/oregon-winery-ai-legal-fight.html Subscription required.
- [3] United States District Court for the District of Oregon, Couvrette v. Wisnovsky, No. 1:21-cv-00157-CL, Opinion and Order (ECF No. 215), CourtListener, December 12, 2025. https://storage.courtlistener.com/recap/gov.uscourts.ord.158388/gov.uscourts.ord.158388.215.0.pdf
- [4] United States District Court for the District of Oregon, Couvrette v. Wisnovsky, No. 1:21-cv-00157-CL, Opinion and Order (ECF No. 225), GovInfo, March 23, 2026. https://www.govinfo.gov/content/pkg/USCOURTS-ord-1_21-cv-00157/pdf/USCOURTS-ord-1_21-cv-00157-1.pdf
This article is provided for informational purposes by DP3. It is not legal advice. The case details referenced are drawn from published reporting and public court records.