Ex-NPR Host Claims Google Stole His Voice For NotebookLM AI Podcast Tool

A public radio mainstay is taking on Big Tech over a voice he says sounds a little too familiar. David Greene, the longtime host of NPR's Morning Edition and current co-host of political talk show Left, Right & Center, has filed a lawsuit alleging that Google's AI podcast feature in NotebookLM cloned or closely replicated his voice without permission.

See, NotebookLM, a research and note-taking tool from Google, allows users to upload documents and then generate summaries, study guides, and even conversational "Audio Overviews." Those overviews are essentially auto-generated mini-podcasts: two AI hosts banter back and forth about the source material in a polished, NPR-ish cadence designed to sound natural and engaging.

According to the complaint, Greene began hearing from friends and colleagues who were convinced they'd stumbled onto his latest side project. The male AI voice used in NotebookLM's podcast feature reportedly mirrors not just a generic public-radio tone, but in fact Greene's specific rhythms, phrasing, and vocal tics. The resemblance, he claims, was strong enough that listeners assumed it was actually him.

The lawsuit, filed in Santa Clara County, alleges that Google trained or modeled the AI voice using Greene's past broadcasts without consent or compensation. The filing references forensic audio analysis that allegedly found a significant similarity between Greene's recorded voice and the AI-generated one. For its part, Google has denied the allegation, stating that the NotebookLM voice was created using a paid voice actor and was not trained on Greene's recordings. The company has characterized the lawsuit as baseless.


A similar controversy surfaced last year when Scarlett Johansson objected to a synthetic voice unveiled by OpenAI that many listeners felt sounded strikingly like hers. Johansson publicly criticized the company and said she had previously declined an offer to license her voice. OpenAI denied intentionally copying her and ultimately paused the voice in question. The dispute did not result in a lawsuit, but it amplified concerns about consent and voice rights in the generative AI era.

Zooming out, this isn't just about whether Google borrowed Greene's voice. As AI-generated speech gets more convincing, the distinction between "a generic broadcast voice" and "that's clearly this person" starts to sharpen. Broadcasters build careers on recognizability. If software can approximate that identity closely enough to confuse audiences, courts may have to decide whether resemblance alone is enough to trigger liability.

Greene's case could test those concerns in court. Whether the similarity is coincidence, convergence, or something more deliberate will likely become a battle of experts. Whatever happens, the reality is that in a world where AI can spin up a podcast host in seconds, the question of how much a person owns their own voice is going to have to be answered by the courts sooner or later.

_{Top image by Studio_Iris from Pixabay.}