Journalists Sue Tech Giants Over Voice Model Training

According to the Chicago Sun-Times, nine class-action lawsuits were filed this week in Chicago federal court invoking Illinois' Biometric Information Privacy Act, or BIPA. LegalNewsline and Biometric Update report the complaints were filed by Illinois-based broadcast journalists, podcasters and voice actors and name technology firms including Google, Amazon, Apple, Microsoft, Meta, NVIDIA, ElevenLabs, Adobe and Samsung.

Biometric Update reports that some complaints focus on Google and allege the company built foundational voice models from thousands of hours of recorded speech and extracted biometric "voiceprints" used to power products such as Gemini Live, NotebookLM Audio Overviews, YouTube auto-dubbing, Google Cloud Text-to-Speech and Google Assistant. LegalNewsline lists named plaintiffs including Carol Marin, Phil Rogers, Alison Flowers, Robin Amer, Lindsey Dorcus, Yohance Lacour and Victoria Nassif. The filings seek statutory damages and injunctive relief, Biometric Update reports.

The Chicago Sun-Times quotes Ross Kimbarovsky, an attorney with the Chicago firm Loevy & Loevy: "What we are seeing is an illegal and unethical exploitation of talent on a massive scale, and one of the largest violations of biometric privacy ever committed." The Sun-Times also places the new complaints in the longer history of BIPA litigation, including the class-action settlements that followed earlier facial-recognition and fingerprint disputes.

The complaints hinge on two technical-legal claims: that voices constitute a biometric identifier under BIPA, and that modern speech models effectively encode extractable, persistent biometric features. Biometric Update characterizes the lawsuits as alleging that voice models internalize voiceprints in model parameters so the biometric data cannot be deleted simply by removing files. Industry observers have documented that large text-to-speech and voice-cloning systems can reproduce speaker characteristics from training data; that technical fact supports the plaintiffs assertion that models can replicate identifiable voice signatures, though the legal question is whether that process constitutes the collection and storage of biometric identifiers as defined by statute.

Editorial analysis: Illinois' BIPA has produced extensive litigation over the past decade, most notably a $650 million settlement with Meta over facial recognition, the Chicago Sun-Times reports. Public reporting frames the current suits as a next wave of BIPA cases focused on AI-derived biometric outputs rather than discrete device-captured fingerprints or faceprints. For practitioners, the cases underscore how privacy statutes originally drafted for sensors and cameras are being applied to machine-learned representations and downstream product features.

For practitioners: monitor several concrete indicators in the litigation and product ecosystem.

• •Rulings on whether model parameters or generated audio constitute "biometric identifiers" under BIPA; lower-court interpretations will set precedent.
• •Motions to dismiss and class-certification briefing, which LegalNewsline and Biometric Update note are likely early milestones.
• •Any public disclosures or changes in vendor retention/consent policies by companies named in the complaints.
• •Regulatory or legislative responses at the state or federal level addressing biometric data and AI-generated content.

These suits combine technical claims about how models encode speaker characteristics with statutory privacy protections. The complaints, as reported, present a fact pattern that will force courts to address whether training from publicly available speech crosses the statutory line that BIPA draws around biometric data.