Radio Host David Greene Says Google's NotebookLM Tool Stole His Voice

Veteran NPR host David Greene has publicly accused Google of cloning his voice without consent for its viral NotebookLM Audio Overview feature. The allegation throws a sharp spotlight on a question every content creator and business owner should be asking: who actually owns your voice, your brand, and your identity in the age of generative AI?

What Exactly Happened Between David Greene and Google NotebookLM?

David Greene spent years as co-host of NPR's Morning Edition, one of the most widely heard radio programs in the United States. His voice is instantly recognizable to millions of listeners. When Google launched NotebookLM's Audio Overview feature in late 2024, the tool quickly went viral for its ability to turn uploaded documents into surprisingly natural podcast-style conversations between two AI-generated hosts.

Greene and numerous listeners noticed something unsettling: one of the AI-generated voices bore a striking resemblance to Greene's own vocal cadence, tone, and delivery style. Greene stated publicly that neither he nor his representatives were ever contacted by Google for permission, licensing, or compensation. Google denied the voice was modeled after any specific individual, maintaining that its voices are fully synthetic and not intentional replicas of real people.

Regardless of intent, the incident crystallized a fear that has been building across media, entertainment, and business: AI systems trained on vast datasets of public audio can reproduce someone's likeness closely enough to cause real harm, whether or not a company admits to deliberate cloning.

Why Should Business Owners Care About AI Voice Cloning?

This is not just a celebrity problem. If you run a business, produce content, host a podcast, or record training videos, your voice and brand identity are commercial assets. The Greene case illustrates how quickly those assets can be replicated, redistributed, or devalued without your knowledge.

• Right of publicity at risk: Many U.S. states protect individuals against unauthorized commercial use of their likeness, including their voice. AI-generated lookalike voices may violate these protections.
• Training data opacity: Most AI companies do not disclose exactly which audio recordings are used to train voice synthesis models, leaving creators in the dark about whether their work has been consumed.
• No federal AI voice law yet: While several states have introduced or passed legislation targeting AI deepfakes and voice replicas, there is no comprehensive federal standard, creating a patchwork of protections.
• Brand trust erosion: If customers or audiences hear an AI voice that sounds like yours promoting products or ideas you never endorsed, the resulting confusion can damage hard-earned brand credibility.
• Precedent from entertainment strikes: The 2023 SAG-AFTRA and WGA strikes placed AI replication of human performances at the center of labor negotiations, establishing that voice and likeness protections are a mainstream business concern, not a niche legal theory.

How Does NotebookLM's Audio Overview Actually Work?

Google's NotebookLM allows users to upload documents such as PDFs, articles, and notes, then generates a conversational audio summary featuring two AI hosts who discuss the material. The technology relies on large language models for script generation and advanced text-to-speech synthesis for voice production. Google says the voices are built from synthetic data, not cloned from identifiable individuals.

However, modern text-to-speech models are trained on enormous corpuses of recorded speech. Even if a final output voice is not a one-to-one clone, the aggregate influence of thousands of hours of real human speech inevitably shapes the result. Critics argue this makes the distinction between "synthetic" and "cloned" more semantic than substantive. When the output is indistinguishable from a real person's voice to trained listeners and colleagues, the practical impact is the same.

Key insight: The legal and ethical debate is no longer about whether AI can replicate a human voice. It clearly can. The real question is whether existing intellectual property and publicity frameworks are strong enough to protect individuals and businesses before the damage is done, or whether we are building guardrails after the car has already left the road.

💡 DID YOU KNOW?

Mewayz replaces 8+ business tools in one platform

CRM · Invoicing · HR · Projects · Booking · eCommerce · POS · Analytics. Free forever plan available.

Start Free →

What Can Businesses Do to Protect Their Voice and Brand Identity?

Waiting for legislation is not a strategy. Forward-thinking businesses are already taking concrete steps to safeguard their identities. Start by auditing every piece of audio and video content your company has published. Understand where your recordings live, who has access, and what licensing terms govern their use.

Register trademarks for distinctive brand elements including taglines, jingles, and brand voice guidelines. Consider adding explicit AI-restriction clauses to contracts with media platforms, podcast hosts, and content distributors. If you discover an AI-generated voice that closely mimics your own or your company's spokesperson, document it immediately and consult legal counsel familiar with right-of-publicity law in your jurisdiction.

Most importantly, centralize your brand assets and communications in a platform you control. When your customer interactions, content, marketing, and operations run through a single system with clear audit trails, you have far greater visibility into how your brand is being represented and far more leverage if someone misuses it.

Frequently Asked Questions

Is AI voice cloning illegal?

It depends on your jurisdiction. Several U.S. states have right-of-publicity laws that protect against unauthorized use of a person's voice for commercial purposes. Tennessee's ELVIS Act and proposed federal legislation like the NO FAKES Act specifically target AI-generated replicas. However, enforcement is inconsistent, and many regions lack clear legal frameworks. If you believe your voice has been cloned without consent, consult an attorney who specializes in intellectual property or entertainment law.

Did Google admit to using David Greene's voice in NotebookLM?

No. Google has maintained that NotebookLM's Audio Overview voices are fully synthetic and not modeled after any specific person. However, Greene and multiple listeners independently identified a strong resemblance to his recognizable vocal style. The disagreement underscores a broader transparency gap: companies rarely disclose the specific training data behind their AI models, making independent verification nearly impossible.

How can I check if my voice or content has been used to train an AI model?

Currently, there is no simple public tool that lets individuals search AI training datasets for their own voice or content. Some organizations like Have I Been Trained allow visual artists to check image datasets, but equivalent tools for audio are still limited. The best proactive measures are monitoring AI-generated content in your industry, setting up alerts for your name and brand, and keeping detailed records of all original content you produce so you can demonstrate ownership if a dispute arises.

The David Greene case is a warning shot. Whether you are a solo creator or managing a 50-person team, your voice and your brand are assets worth protecting. The businesses that move fastest to centralize their operations, document their intellectual property, and control their customer-facing identity will be the ones best positioned as AI regulation catches up to AI capability.

Ready to take control of your business operations and brand in one place? Mewayz gives you 207 integrated modules to run your entire business, from CRM and marketing to project management and client communications, all under one roof. Join 138,000+ users who already trust the platform. Start your free trial at app.mewayz.com and put your business on solid ground.