Experts Warn Against ElevenLabs AI Voice Clone in Court
TL;DR. Legal teams and tech experts oppose using an ElevenLabs AI-generated voice of a murder victim in a Utah court case. - Prosecutors created a synthetic voice from old audio and testimony for a 1985 murder sentencing. - Defense experts argue AI voices can mislead jurors due to perceived humanlike qualities and potential inaccuracies. - A judge will decide if the AI voice can be presented as evidence in the upcoming February 2027 sentencing.
- Prosecutors in Utah seek to use an AI-generated voice of a murder victim, Joyce Yost, in a 1985 capital murder case.
- The voice clone was created using ElevenLabs software, based on actual audio and testimony from the victim.
- Defense experts, including computer science professor Sameer Patil, warn that AI-generated voices can be misleading, as people tend to imbue AI content with human characteristics.
- The primary concern is the potential for jurors to misinterpret or over-rely on a synthetic voice, impacting the fairness of the legal process.
- A judge's decision on the admissibility of the AI voice is pending for November, with the sentencing scheduled for February 2027.