Judge denies Meta summary judgment in privacy class action over unauthorized voiceprint collection
Judge denies Meta summary judgment in privacy class action over unauthorized voiceprint collection
Publish Date: 2026-06-25 15:32:00
Source Domain: www.courthousenews.com
SAN FRANCISCO (CN) — A federal judge unsealed an order Thursday advancing privacy claims against Meta brought by a class of Illinois consumers who say Meta took their voiceprints without permission in violation of an Illinois biometric privacy law.
In the order, originally filed on May 20, U.S. District Judge Susan Illston denied Meta’s motion for summary judgment, finding the plaintiff provided enough evidence to create a genuine issue of material fact as to whether Meta collected her voiceprint.
The judge, a Bill Clinton appointee, concluded the plaintiff had shown there were disputed facts regarding whether Meta had collected the plaintiff’s voice recording in a way the tech company could identify the plaintiff. She also found that there was enough evidence that Meta possesses the technology to process a voice recording and link it to a user’s account and personally identifiable data, such as name, birthday and address, that Meta associates with their account.
“Today’s decision need not and does not attempt to precisely delineate at what point voice data transforms from a ‘mere voice recording,’ as Meta puts it, into a ‘voiceprint’ under BIPA,” she wrote in the 12-page order. “For today, it is enough that there is a dispute of material fact regarding whether Meta has collected biometric data that is capable of identifying an individual using technology Meta possesses.”
The named plaintiff, Natalie Delgado, is an Illinois citizen who says Meta took her voiceprint — a digital representation of a person’s unique voice characteristics — without complying with the requirements of Illinois’ Biometric Information Privacy Act.
Delgado claims Meta uses the audio input into Facebook or Messenger to create encoded data of the speaker’s voice, and that data is then processed with an acoustical model that is then trained and further refined using the voice of a particular speaker, such that the acoustical model can be used to recognize that…
