NPR Host Sues Google, Claims AI Used His Voice Without Permission

A lawsuit filed in California Superior Court alleges that Google unlawfully used the voice of former NPR host David Greene to train its AI-powered research assistant, NotebookLM. Greene, who previously hosted NPR’s Morning Edition and currently hosts KCRW’s Left, Right & Center, claims Google replicated his voice without permission or compensation.

The core of the dispute centers around NotebookLM’s audio overview feature, introduced in late 2024. This functionality allows users to generate AI-narrated podcasts from uploaded text files. Greene says he was alerted by listeners who noticed a striking similarity between the AI-generated voice and his own. He subsequently commissioned a forensic voice analysis, which reportedly indicated a 53%-60% confidence level that his voice was used in the software’s training data.

According to the lawsuit, Google “used Mr. Greene’s voice without authorization and then used those stolen copies to develop, train, and refine its AI broadcasting product, NotebookLM.” The complaint doesn’t detail *how* Google allegedly obtained access to Greene’s voice, but implies it was done without his consent.

Google vehemently denies the allegations. A spokesperson told CNET, “These allegations are baseless. The sound of the male voice in NotebookLM’s Audio Overviews is based on a paid professional actor Google hired.” However, the company has not publicly identified the actor in question, leaving open the question of how the AI voice arrived at its current form.

The Broader Context: AI Voice Cloning and Intellectual Property

Greene’s lawsuit arrives at a critical juncture for the rapidly evolving field of AI voice technology. The ability to convincingly replicate human voices has advanced dramatically in recent years, raising complex legal and ethical questions about intellectual property, consent, and the potential for misuse. This isn’t the first instance of concern surrounding AI-generated voices mimicking public figures.

In 2024, actress Scarlett Johansson expressed concerns about an OpenAI voice that sounded remarkably like her own, prompting the company to remove the voice. More recently, companies like ElevenLabs have begun licensing voice rights from celebrities, including Matthew McConaughey and Michael Caine, establishing a framework for legal and compensated use of an individual’s vocal identity. These licensing deals suggest an industry recognition that voices, like likenesses, possess inherent value and require protection.

How NotebookLM Works and the Technical Challenges of Voice Attribution

NotebookLM is designed as an AI-powered research assistant. Users upload documents, and the system can summarize information, answer questions, and, with the audio overview feature, generate spoken-word content. The technology relies on large language models (LLMs) trained on massive datasets of text and audio. The process of creating a realistic AI voice involves training a model to map text to speech, capturing nuances in pronunciation, intonation, and pacing.

Determining whether an AI voice is based on a specific individual’s voice is a complex technical challenge. While forensic voice analysis can identify similarities, achieving definitive proof of replication is difficult. The 53%-60% confidence rating reported in Greene’s lawsuit, while suggestive, falls short of conclusive evidence. Factors like vocal training, speaking style, and the inherent variability in human speech can complicate the analysis. The lawsuit highlights the difficulty in establishing a clear line between legitimate imitation and unlawful appropriation in the age of AI.

Implications for Voice Actors and Content Creators

The outcome of Greene’s lawsuit could have significant implications for voice actors, broadcasters, and other content creators. If Google is found to have unlawfully used Greene’s voice, it could set a precedent for holding AI developers accountable for unauthorized voice replication. This could lead to stricter regulations governing the use of voice data in AI training and a greater emphasis on obtaining explicit consent from individuals whose voices are used.

The case also underscores the need for clearer legal frameworks surrounding AI-generated content. Current copyright laws are often ill-equipped to address the unique challenges posed by AI, particularly when it comes to protecting intangible assets like voice and likeness. The legal landscape is evolving, but the fundamental question remains: who owns the rights to a voice in the age of artificial intelligence?

The lawsuit is ongoing, and Google maintains its denial of wrongdoing. However, the case serves as a stark reminder of the potential risks and ethical dilemmas associated with AI voice technology and the importance of protecting the intellectual property rights of individuals in a rapidly changing technological landscape.