Lawsuit tracker
nine coordinated suits against Adobe, Alphabet/Google, Amazon, Apple, ElevenLabs, Meta, Microsoft, NVIDIA and Samsung · ActiveIllinois Voiceprint BIPA Cases
Nine lawsuits, one statute, nine of the biggest names in tech. Illinois journalists and voice actors say Adobe, Google, Amazon, Apple, ElevenLabs, Meta, Microsoft, Nvidia and Samsung extracted their voiceprints without consent — and under Illinois law, fair use isn't a defense to that claim.
Key facts
- In mid-May 2026, seven Illinois voice professionals filed nine coordinated class actions — one per defendant — against Adobe, Alphabet/Google, Amazon, Apple, ElevenLabs, Meta, Microsoft, Nvidia and Samsung.
- Named plaintiffs include broadcast journalists Carol Marin and Phil Rogers, podcaster-journalists Robin Amer, Yohance Lacour and Alison Flowers, and audiobook narrators and voice actors Lindsay Dorcus and Victoria Nassif.
- The suits allege violations of the Illinois Biometric Information Privacy Act (BIPA): collecting, storing and profiting from voiceprints — mathematical representations of pitch, timbre and resonance — without written notice, a signed release, or a public retention policy.
- BIPA provides statutory damages of $1,000 to $5,000 per violation, and the complaints cite the defendants’ own consent flows for customer-uploaded voices, plus prior nine- and ten-figure biometric settlements involving Meta and Google, as evidence the companies knew consent was required.
- The suits are only weeks old as of July 2026. No defendant responses, consolidation orders or rulings have been publicly reported.
Nine suits, one law firm, one theory
Loevy + Loevy filed nine separate class actions in Illinois during the week of May 11, 2026 — one complaint per defendant, all built on the same legal theory. Seven named plaintiffs appear across the suits in varying combinations; per the firm’s announcement, Alison Flowers appears on all of them except the suits against Amazon and Apple.
The plaintiffs are not anonymous class representatives. Carol Marin and Phil Rogers are Emmy- and Peabody-recognized Chicago broadcast journalists. Robin Amer, Yohance Lacour and Alison Flowers are journalist-podcasters. Lindsay Dorcus and Victoria Nassif are audiobook narrators and voice actors. Their public-facing careers are exactly what put decades of their recorded voices within reach of a web scrape.
The claim: a privacy statute, not a copyright one
These suits skip copyright entirely and go straight at the Illinois Biometric Information Privacy Act, in force since 2008. BIPA requires written notice of collection purpose and retention schedule, a signed release, and a publicly available retention policy before anyone collects a person's biometric identifier — a defined category that includes voiceprints.
The complaints allege the defendants extracted voiceprints, described as mathematical representations of a speaker’s pitch, timbre and resonance, from publicly available audio like broadcasts, podcasts and audiobooks, then used them to train commercial voice and text-to-speech models. None of the plaintiffs gave written consent for that specific use, and none of the defendants, the suits allege, published a retention policy covering it.
Why "the audio was public" doesn’t work here
The defense that has carried real weight in copyright cases — that training on lawfully accessible, public content is fair use — has no equivalent under BIPA. The statute doesn’t ask whether a use was transformative or whether it harmed a market. It asks whether a company collected a biometric identifier without the specific written consent the law requires. Publicly broadcasting your voice on the radio is not the same, under BIPA, as consenting to have it fingerprinted and fed into a training set.
The complaints reinforce that point by pointing to the defendants’ own behavior elsewhere: several of these same companies already require explicit consent flows when customers upload their own voices for cloning features. The suits argue that pattern shows the companies understood consent was legally required — they just didn’t apply it to voices they scraped rather than voices users uploaded.
What this means for voice-data provenance
For anyone building or buying voice or speech-to-speech training data, this is a distinct compliance track from copyright clearance, and it can’t be satisfied the same way. A dataset can be fully copyright-cleared — public domain, licensed, or otherwise fair game — and still expose a buyer to BIPA liability if any speaker in it was recorded in Illinois and never gave written biometric consent.
Statutory damages of $1,000 to $5,000 per violation, assessed per person rather than per work, make voice data a different order of risk than text or image data at scale. Any vendor supplying speech, voice-cloning or speaker-identification datasets should now be asked directly whether Illinois residents are represented in the corpus and whether biometric consent was obtained — not just whether the underlying recordings were licensed.
What to watch
- Whether any of the nine defendants move to dismiss, and on what grounds — likely including standing, extraterritoriality, and whether voiceprints derived from broadcast audio qualify as BIPA biometric identifiers.
- Whether the nine suits are consolidated for pretrial proceedings given their shared plaintiffs, law firm and factual theory.
- Whether other Illinois voice professionals or unions file additional suits, or whether the case prompts similar filings in other biometric-privacy states.
- Whether voice-model companies begin publishing BIPA-compliant consent and retention policies in response, as several already do for user-uploaded voice clones.
Sources
See something wrong? Send a correction.
Jaeden Schafer
Jaeden Schafer is the founder of fiund and host of the AI Chat podcast. He covers the training-data market and the lawsuits shaping it.
More case files
Want data that clears this in diligence?
Whether you're building a model or sitting on an archive, the first conversation is short and specific.
Send a brief