He Spent Decades Perfecting His Voice. Now He Says Google Cloned It.

February 15, 2026

David Greene built his career on his voice.

Warm. Measured. Conversational.

For years, he hosted NPR’s “Morning Edition” and later co-hosted political podcasts. Millions woke up to that baritone.

Then a former colleague emailed him:

“Did you license your voice to Google? It sounds exactly like you.”

He hadn’t.

🤖 The Tool at the Center

The product:

Google’s NotebookLM.

It can generate AI podcasts on demand — complete with two virtual hosts engaging in casual banter.

When Greene listened, he says he froze.

The cadence.
The pauses.
The subtle “uhhs.”
Even the tonal warmth.

“I felt like I was listening to myself,” he said.

⚖️ The Lawsuit

Greene is now suing Google.

He claims the company built a product that replicated his voice without consent — allowing users to generate speech that sounds like him saying things he never said.

Google denies it.

A spokesperson says the voice is based on a paid professional actor — not Greene.

Now the case may hinge on a question courts are just beginning to confront:

How similar does an AI-generated voice need to be to count as infringement?

🎧 The Bigger AI Voice Problem

This isn’t an isolated dispute.

We’ve already seen:

• Political deepfake voice clips
• AI voices resembling Scarlett Johansson
• Celebrity “voicefakes” in scam ads

The line between “inspired by” and “copied from” is getting blurry.

And AI doesn’t need a 100% match.

Sometimes 70% is enough for listeners to assume it’s real.

Courts may need to decide:

  1. Would ordinary listeners believe it’s Greene?

  2. Is his voice distinctive enough to be protectable?

  3. Does similarity alone equal harm?

Past cases exist — like Bette Midler successfully suing Ford in the 1980s over a voice imitation in a commercial.

But AI complicates everything.

It’s not just imitation.

It’s probabilistic synthesis.

💔 Why It Feels Personal

For Greene, this isn’t about missed money.

It’s about identity.

“My voice is the most important part of who I am,” he said.

He worries about an AI version of “him” lending credibility to misinformation or low-quality content.

When your profession is trust…

A synthetic twin feels existential.

🌐 The Larger Conflict

This case sits at the center of a much bigger tension:

Individual creator rights
vs
Mass-scale AI training on public data

Models are trained on enormous libraries of speech and text — often without explicit consent from the humans who created them.

Is that fair use?

Or digital appropriation?

🚨 What Happens Next?

If a jury decides the resemblance is strong enough, Greene could win.

If they rule the voice is simply a generic “anchorperson archetype” learned from massive datasets, Google could prevail.

Either way, this won’t be the last case.

AI voice generation is exploding.

And the legal framework is still catching up.

This isn’t just about one journalist.

It’s about a future where:

Your face.
Your voice.
Your tone.
Your identity.

Can be statistically reconstructed.

The question is no longer whether AI can sound like you.

It’s whether it should — without asking.