Deepfake Video Has Snuck into the Courtroom (Video)

This is a 27 December 2025 published video from host Steve Lehto of Lehto’s Law. 

In 2025, the legal system has reached an “existential moment” as AI-generated deepfakes have begun appearing in courtrooms as purportedly authentic evidence. While some instances have been detected, experts warn that many others may have already gone unnoticed. 

Mendones v. Cushman & Wakefield (California): 

In September 2025, an Alameda County judge threw out this civil case and recommended sanctions against the plaintiffs for submitting a deepfake video of witness testimony. The judge detected the fraud due to “telltale glitches” like repeated eye blinks and looped video, though experts noted modern AI can now solve these flaws, making such detection “lucky”.

Experts like Brian Long (CEO of Adaptive Security) argue that the legal system’s traditional reliance on “seeing is believing” is no longer viable. Courts are being urged to shift toward “source-centric” authentication, focusing on metadata, source stamping, and verified chains of custody.

In late 2025, reports and judicial panels proposed national standards and specialized training for judges and jurors to evaluate AI-enhanced footage. New resources, such as “bench cards” from the National Center for State Courts (NCSC), help judges navigate the admissibility of both acknowledged (disclosed) and unacknowledged AI evidence. 

https://youtu.be/Bxa1M853i6g

AI can generate legal briefs with fake case citations (known as “hallucinations”), and this has resulted in significant sanctions for numerous lawyers across the United States and in other countries. Courts universally emphasize that it is an attorney’s non-delegable duty to verify every citation they submit, regardless of the tools used.

The primary cause of the issue is lawyers failing to personally read and verify the AI’s output against actual legal databases like Westlaw or LexisNexis before filing documents with the court. Many attorneys have admitted they relied on AI for efficiency without understanding its limitations, mistakenly believing the AI would provide accurate legal research.

Lawyers who have submitted briefs with unverified, AI-generated citations have faced serious professional and financial penalties.

Courts have imposed fines ranging from thousands of dollars on individual attorneys and law firms. Attorneys have been referred to their state bar’s disciplinary committees for investigations into professional misconduct and potential discipline. Some sanctions have included requirements for lawyers to complete specific ethics training regarding AI use. The cases are often highly publicized, leading to significant reputational harm. In some instances, the underlying legal case or appeal has been dismissed due to the procedural errors and failure to provide valid authority.