Where Class Actions Are Going in 2027: AI Training, Voice Cloning, and the BIPA Wave
The next wave of consumer class actions is forming around AI. Here is what is already in pre-trial and what consumers should track.
Class action trends shift in three-to-four-year cycles. Data breach cases peaked from 2018-2024. BIPA biometric cases are at peak now. The next wave forming for 2026-2028 is AI training data, voice and likeness cloning, and the second-generation BIPA cases that follow facial geometry into a much broader range of biometric features.
AI training data cases
Multiple authors and rights holders have sued OpenAI, Anthropic, Meta, Microsoft, and Google for training on copyrighted works without licensing. Most current cases are filed by named authors as direct copyright cases, not class actions. But three filed in late 2025 and early 2026 are seeking class certification for any author whose work was included in widely-used training datasets such as Common Crawl, Books3, and the LibGen corpus.
If certified, payouts to individual authors would be modest (per-work statutory damages distributed pro rata across millions of works) but the broader effect would be the licensing requirements that the cases force into industry practice.
Consumer class actions for AI training (where the plaintiff is an end-user, not a rights holder) are not yet on file. The likely vector is image and video content uploaded to social platforms that was scraped into training datasets, similar to the BIPA model.
Voice and likeness cloning
The Hollywood actor and SAG-AFTRA cases targeting voice clone services in 2025-2026 are progressing through commercial litigation, not consumer class actions. But two related consumer cases are early in the docket:
A 2026 Massachusetts case against a voice-clone consumer app alleges the app used user-uploaded voice samples to train its general model without disclosure. A California case from late 2025 targets a video-clone app for similar conduct with image data. Both are early class certification motions.
If certified, these will be the first true consumer class actions in the AI clone space, and the per-claimant payouts will follow the BIPA precedent: roughly $50-$150 per claimant.
Second-generation BIPA expansion
The Snap, Meta, Google, and Clearview BIPA cases covered facial geometry. Newer cases target voice biometrics, gait analysis, eye-tracking, and emotion-recognition systems. The Illinois statute is the most plaintiff-friendly biometric privacy law in the US, and its definition of "biometric identifier" is broad enough to capture all of these.
Three cases to watch: a 2026 case against a workplace surveillance vendor for facial emotion analysis without consent, a 2025 case against a fitness app for body geometry tracking, and the ongoing investigations into voice-print systems used by some banks for customer authentication.
What this means for consumers in 2026-2027
File for any active settlement you qualify for now. Most cases that will affect you over the next two years are still in pre-trial. The settlement landscape in 2027-2028 will be heavier on biometric privacy, AI training, and platform liability.
Two practical actions: keep records of which apps and services you used during specific years, and check Have I Been Pwned and the BIPA case trackers every six months. The eligibility criteria for these cases are usually "used the service in state X during year range Y", so accurate personal records matter.