The big picture: Judges are already facing a new challenge in courtrooms as AI-generated documentation worms its way into filings. Now that voice cloning and video deepfakes have become more convincing, courts must scrutinize video and audio testimony carefully. This unprecedented extra vetting is only going to bog down an already complicated legal system.
A California housing dispute is getting media attention over allegations that lawyers presented a deepfake video as witness testimony. NBC News reports that Judge Victoria Kolakowski became suspicious after the supposed witness showed signs that something was not right, including a monotone voice, fuzzy facial features, and repeated facial expressions. Judge Kolakowski soon realized that the clip had the hallmarks of generative AI.
The case, Mendones v. Cushman & Wakefield, may be one of the earliest instances of lawyers submitting deepfake video as authentic testimony. Judge Kolakowski dismissed the case in September and denied a request for reconsideration in early November. Legal experts warn that the incident highlights a broader threat, as AI-generated evidence is increasingly flooding courtrooms and compromising the judicial system in unprecedented ways.
Mendones v. Cushman & Wakefield is not the first time courts have caught generative AI in filings and evidence. In February, a judge fined a lawyer $15,000 for submitting filings with fake AI-generated case citations. Instances of AI misuse will continue to accumulate as companies, including law firms, integrate it into their practices without proper oversight, and judges are already grappling with the implications.
Judge Scott Schlegel of Louisiana, who actually supports judicial AI adoption, cautions that plaintiffs could use AI-cloned voices to generate threatening recordings, falsely affecting decisions in restraining order cases. Similarly, Judge Erica Yew pointed out that forged documents can easily enter official records, challenging traditional trust in public filings. Despite the mounting instances and increasing concerns, the legal system has yet to establish a centralized system to track these incidents.
Courts are experimenting with guidance to address AI-generated content, but there are no formal protocols. The National Center for State Courts and the Thomson Reuters Institute classify deepfakes as "unacknowledged AI evidence," offering judges checklists to verify origin, access, and alteration. Of course, this requires judicial review that goes beyond the norm.
Brian Long, CEO and Co-Founder of Adaptive Security, says the problem is only getting more complicated.
"The hard truth is that next-gen AI makes these fakes incredibly convincing, and detection tools are not keeping up. That means law firms need new processes fast," Long told TechSpot. "Always verify audio and video through a second channel, request the original source files, and confirm live with the real individual whenever possible. The earlier you can spot an impersonation attempt, the safer your clients will be.
Judges are now navigating a new frontier where traditional evidence standards collide with generative AI. Courts are only beginning to adapt, and experts caution that without stronger protocols, dockets could soon start admitting material that appears real but is entirely fabricated. Some judges are already calling for centralized tracking and more precise guidance. However, until the AI industry sees some real regulation, it's the wild west out there.
Judge dismisses lawsuit twice due to alleged deepfake video testimony
