June 1, 2026 (in 3 days): New York: 22 NYCRR Part 161 takes effect, system-wide AI policy for all UCS courts

Deutsche Bank National Trust Co. v. LeTennier

New York Supreme Court, Appellate Division, Third Department · N.Y. App. Div. 3d Dep't · New York bar guidance

Court sanction

Verified May 5, 2026

Citation
Deutsche Bank Natl. Trust Co. v. LeTennier, --- N.Y.S.3d ----, 2026 WL 53120, No. CV-23-0713 (N.Y. App. Div. 3d Dep't Jan. 8, 2026)
Decided
January 8, 2026

Summary

Defendant's appellate counsel Joshua A. Douglass submitted five filings during a mortgage foreclosure appeal that contained no fewer than 23 fabricated legal authorities, along with multiple misrepresentations of holdings from real cases. After plaintiff identified the fake citations, Douglass conceded at oral argument that he had used generative AI to prepare the papers and estimated 90% of his citations were accurate. Justice Fisher, writing for a unanimous panel (Aarons, J.P., Reynolds Fitzgerald, Ceresia, Fisher and McShan, JJ.), found that more than half of the fabricated authorities were filed after counsel was on notice of the problem, and that defense counsel "doubled down" rather than taking remedial measures.

AI tool:
Unspecified generative AI
Sanction amount:
$10,000
This case summary is informational only. Verify the underlying opinion or order against the primary source before relying on it in any filing or client matter.

What sanction did the court impose?

Underlying orders affirmed with costs. The court imposed $5,000 on Douglass for the fabricated authorities and an additional $2,500 each on Douglass and on defendant Jean LeTennier for prosecuting a frivolous appeal, for a combined sanction of $10,000 payable within 60 days under 22 NYCRR 130-1.1. Counsel's portion is payable to the Lawyers' Fund for Client Protection; the party's portion to the Commissioner of Taxation and Finance.

Why does Deutsche Bank National Trust Co. v. LeTennier matter for law firms using AI?

LeTennier is the first New York appellate decision to impose sanctions for misuse of generative AI, and it sets out a multi-factor framework, volume of fabrications, persistence after notice, presence of fake quotations, remorse, and impact on proceedings, that other panels are likely to follow. The case is a useful internal training artifact for two reasons. First, the volume of fabrications is uncommonly high (23 fake authorities across five filings), and the panel was most troubled by the fact that more than half were filed after opposing counsel had flagged the problem, an escalation pattern that supervising attorneys should be alert to when reviewing AI-assisted work. Second, the unusual split sanction, $7,500 against counsel and $2,500 against the pro se-leaning client whose document metadata appeared on filings counsel signed, illustrates that courts will look past the signature block when filings appear to originate with the litigant rather than the attorney of record.

Sources

Primary sources

Further reading

Source PDF is a Westlaw printout mirrored from the Damien Charlotin hallucination database. We are working to add the underlying court docket (PACER, CourtListener, or court website) as a second source.