Skip to content

Courts on AI / Fabricated or misquoted citations

United States v. Karnezis

United States District Court for the District of Oregon · Federal district courts · OR · · 2026 WL 2212298 · No. 3:23-cr-00067-IM · order

The District of Oregon partially suppresses statements in a criminal case. In resolving the Miranda arguments, the court notes that defense briefing supplied a quotation that does not appear in Smith and admonishes counsel, while recording counsel’s statement that the mistake was not caused by generative AI.

Disposition
The court granted in part a suppression motion and admonished Williams's counsel for a non-existent quotation in briefing.
Recorded conduct
Defense counsel submitted a non-existent quotation from a Ninth Circuit Miranda case in suppression briefing. — outcome: warning; actor: lawyer
Canonical record
Incident tracker entry — verified outcome, regulator disposition and related rules

What the court wrote

Verbatim from the court's document (public domain). Ellipses mark omitted text; nothing is paraphrased.

Perhaps recognizing that Smith does not support his argument, Williams resorts to providing a non-existent quotation from the case: “The Ninth Circuit held that officers cannot ‘prevent a suspect from making an unambiguous request for counsel by cutting him off.’ ” Reply, ECF 172 at 9. This Court admonishes Williams's counsel for misquoting Smith, but at the same time, this Court does not endorse Agent Pasquale's interruption of Williams. ... At the hearing, Williams's counsel stated that this error was not a result of using generative AI but rather an inadvertent mistake. This Court reminds Williams's counsel that he is responsible for not only his own contributions but also his client's contributions to any of his submissions.

Authorities cited in the passage: United States v. Smith, 860 F.2d 1533 (9th Cir. 1988) · McNeil v. Wisconsin, 501 U.S. 171 (1991) · United States v. Rodriguez, 518 F.3d 1072 (9th Cir. 2008) · Miranda v. Arizona, 384 U.S. 436 (1966)

compiled by the Cognesio LLP editorial teammethodas of 2026-09-07v0.1.1coding CC BY 4.0JSON Hugging Face GitHub