Deepfakes Threaten Court Integrity in Telematic Trials
According to José Montero, CEO of Montero de Cisneros Abogados, the issue lies in the lack of tools to verify whether the voice heard in court is genuinely human. “Until now, deepfakes were presented as evidence—an audio or video file that an expert could examine at their leisure,” he explains. “Now, they could be used as a live witness, with a third party dictating responses in real-time, leaving the judge to assess the credibility of someone who, technically, isn’t testifying.”
Identity, Not Signal, Verified
The 2023 procedural reform standardized videoconferencing in Spanish courts, requiring declarants to connect from a secure judicial office or access point. However, this safeguard only verifies the individual’s identity at the start of the session, not the ongoing integrity of the audio or video feed.
Montero highlights that the risk is highest in cases where witnesses or experts testify from abroad, or when connections are made from private locations outside the control of court officials. “In these scenarios, intermediary software could alter the voice, correct gestures, or feed responses to the witness, and no one in the courtroom would be aware,” he notes.
Real-World Precedents and Legal Gaps
This is not a hypothetical concern. In 2024, British firm Arup transferred $25 million after an employee participated in a video call where all participants were AI-generated impersonations of company executives. Cybersecurity firm Kaspersky ranks real-time deepfake manipulation during video calls among the top threats for 2026. “If it worked to drain a multinational’s funds, it could work to win a lawsuit,” Montero warns.
Read Also: Elevator Installations in Spanish Communities Impact Property Rights
The legal system currently lacks clear provisions to address this threat. Perjury laws target witnesses who lie, while fraudulent litigation requires proof of financial gain and intent to deceive the judge. Neither accounts for individuals whose cloned voices are operated or scripted by someone off-camera.
Even if fraud is discovered, the Civil Procedure Act allows a judgment to be overturned only within five years of its issuance and three months after the fraud is detected, a narrow window that may not suffice.
Proposed Safeguards
To address these vulnerabilities, Montero suggests four measures: cryptographic signatures from cameras and microphones to certify unaltered signals, real-time synthetic voice and image detection integrated into judicial platforms, randomized liveness checks during testimony, and mandatory in-person testimony for critical witnesses when requested, alongside a specific criminal offense for synthetic impersonation in court.
Enhanced Security Measures for Remote Testimonies
Proposed safeguards include cryptographic signatures from cameras and microphones to ensure unaltered signals, similar to existing content authenticity standards. Real-time detection of synthetic voices and images would be integrated into judicial videoconferencing platforms, alerting the court during proceedings.
