• by Wendy Davis , Yesterday

Book narrators, broadcasters, podcasters and others allege in lawsuits filed this week that their voices were wrongly used by companies including Amazon, Apple, Google, Meta, Microsoft and Nvidia to develop artificial intelligence systems or other voice-related technology.

Plaintiffs in the cases include Peabody award winner Carol Marin, Murrow award winner Philip Rogers and podcaster Alison Flowers, among others.

The allegations in all cases are similar, though some specifics differ. For instance, the complaint against Amazon alleges that it built a "global voice-AI business on the voices of real people," while the one against Apple alleges that the company "built its commercial voice synthesis technology by extracting voiceprints from recordings of real people."

Marin and the others typically alleged based on "information and belief" that their voices were used by the tech companies, but also acknowledged that the companies haven't revealed the precise sources of voice training data.

For instance, the complaint against Meta alleges that it "has not publicly disclosed the complete provenance" of training data, but adds that the plaintiffs' voice recordings match a "profile" of training data Meta described in public stock filings and other documents.

The lawsuits, brought in federal court in Illinois, center on two major claims. One is that the companies allegedly violated a state privacy law that requires companies to obtain residents' consent before collecting voiceprints and other biometric data that's used to identify people. The other key claim is that tech companies allegedly violated an Illinois "right of publicity" law that empowers residents to control the commercial use of their voices.

The new lawsuits join dozens of other complaints by authors, news organizations, music companies and other copyright owners  against companies developing artificial intelligence. Most of those cases center on claims that the tech companies infringed copyright by training their large language models on content they didn't own.