Suno Lawsuit: Elvis Estate Sues Over AI Voices + ElevenLabs Archer
Episode 1 of my weekly show on AI voices. The main story: who owns a voice that only sounds like you? After that, a $129 million scam with fake CEO voices, a voice company that shuts down and deletes your data, and what changed at ElevenLabs. Every story with its date and sources.
Disclosure: the ElevenLabs link marked * is an affiliate link. I'm an independent ElevenLabs affiliate.
Chapters
Each timestamp jumps to that part of the transcript below. “Load video” then starts right there.
- 00:00 Who owns a voice that sounds like you?
- 00:50 ElevenLabs' Archer voice and Chris Williamson
- 01:39 Elvis estate sues Suno, Greene v. Google
- 02:32 A study: 41% and what it means for you
- 03:23 A $129M voice scam and Hume shutting down
- 04:33 ElevenLabs: v4 clones, credits, $1 offer
- 05:48 Quick hits and what's coming
This week in 60 seconds
- Podcaster Chris Williamson says ElevenLabs' library voice Archer sounds like him. ElevenLabs says a voice actor made it, and its similarity test scored 0.28, where 0.7 would be a match. Archer is gone from the library.
- Elvis Presley Enterprises and others sued Suno on November 6. The complaint names Suno's Voices feature.
- A California judge let former NPR host David Greene's case over the NotebookLM voice go ahead.
- In a new study, listeners judged a clone and a real recording of the same person to be different people in 41 percent of answers.
- Federal prosecutors in Manhattan charged three men over a $129 million scam with AI-imitated CEO voices.
- Hume shuts down its text-to-speech and EVI voice APIs on November 13 and deletes account data.
- ElevenLabs: professional clones on v4 for every workspace, a one-credit minimum per request and a surcharge for library voices in agents. The $1 first month of Starter runs until November 16.
- Quick hits: HeyGen Voice Turbo is free through the API until November 30, Gemini text-to-speech can remix voices, and ElevenMusic tracks go straight to SoundCloud.
Who owns a voice that only sounds like you?
▶ In the video at 00:00I've cloned my own voice for this channel in ElevenLabs, HeyGen and Voicebox, so this one is personal for me. Four things happened around one question.
The Archer voice at ElevenLabs
October 2026
Chris Williamson hosts one of the biggest podcasts out there. In October he posted a video about a voice called Archer in ElevenLabs' voice library. His friends heard it and said: that's you. He wrote to ElevenLabs.
ElevenLabs' answer, as he reads it out in his video: a voice actor contributed Archer. ElevenLabs says it's checking whether that voice actor did anything fraudulent, but has no reason to believe so. It also ran a similarity test between Archer and his voice. Above 0.7 would mean a match. Archer scored 0.28.
Since then, Archer is gone from the library. Williamson says he doesn't know why. Nobody has sued anyone.
Source: Chris Williamson on YouTube: “The ElevenLabs Situation Is Crazy” (ElevenLabs' reply from 7:20)
Elvis Presley's estate sues Suno
Filed November 6, 2026
On November 6, Elvis Presley Enterprises, together with heirs and publishers of songwriters, sued Suno in federal court in California (Northern District of California, case 3:26-cv-13398). The complaint says Suno lets people produce songs that sound like Elvis singing. It names Suno's Voices feature, the one that lets you make songs with a saved voice. For the likeness claims, it puts the damages at, quote, “believed to exceed $50 million.”
That's an accusation, not a ruling. Suno hasn't had its day in court yet. In another lawsuit, Suno says it “has never allowed prompts for specific artists or copyrighted songs.” The Elvis complaint calls statements like that false.
Sources: Complaint, Elvis Presley Enterprises et al. v. Suno (PDF, hosted by Music Business Worldwide) · Music Business Worldwide: report on the lawsuit, November 10, 2026
Former NPR host vs Google: the NotebookLM case goes on
Order filed October 23, 2026, reported October 29
David Greene, a former NPR host, says the default male voice in Google's NotebookLM is his. Google tried to get the case thrown out. On October 23, a judge in Santa Clara County, California, overruled Google's demurrer, and all four of his claims go on.
That doesn't mean it's his voice. The court hasn't decided that. It means the question is serious enough to fight about.
Source: Radio Ink: Google Must Answer Former NPR Host's AI Voice Claims In Court, October 29, 2026
A new study: 41 percent
November 3, 2026
A study called VIPBench had 1,290 listeners compare voices of 100 public figures, 124,876 judgments in total. In 41 percent of the answers, people said a clone and a real recording of the same person were two different people.
Source: arXiv: VIPBench, a benchmark for voice identity perception (Xudong Tang, Matthew Groh)
Why it matters for creators
So a clone can fail to sound like the person it was made from. And if ElevenLabs is right about Archer, a voice can also sound like someone it was never made from. A similarity score is one answer. What listeners hear is another. And in both lawsuits, what people recognize is a big part of the argument.
If a voice you use sounds like someone famous, I wouldn't publish with it. Here's what I do: I use my own clone or a voice I know the source of, and I keep a note of where each voice came from.
A library voice can also disappear. According to ElevenLabs' help center, if a voice owner stops sharing a voice without a notice period, everyone who saved it loses access right away. With a notice period of 30 days to 2 years, they keep it until the period ends. So if your whole channel runs on one library voice, have a backup.
A $129 million scam with AI-imitated CEO voices
▶ In the video at 03:23Charges announced November 5, 2026
Federal prosecutors in Manhattan announced charges against three men. They're accused of posing as company CEOs and lawyers, using AI to imitate their voices and sometimes their image, and getting companies to wire about $129 million. The prosecutors call the technology “super intelligence.” I'd just call it a convincing phone call.
These are charges, not a verdict. The three men haven't been convicted.
Why it matters for creators: Every YouTuber and podcaster has a public voice, so you can end up as the voice in someone else's fake call. If you run a business, or you're the person people listen to, this is the cheapest protection there is: agree on a code word with your team and your family. And no money moves because of one call. Hang up and call back on a number you already know.
Sources: Associated Press (via KSAT): 3 charged in scheme to use AI to impersonate CEOs and attorneys to steal $129 million · U.S. Attorney's Office, Southern District of New York: press release
Hume shuts down its voice APIs and deletes your data
Shutdown on November 13, 2026, 12:01 a.m. Eastern
Hume, the company behind the Octave voice model, is shutting down its text-to-speech and EVI voice APIs on November 13 at 12:01 in the morning, Eastern time. Hume's notice says any data in your account is permanently deleted after that. Hume posted the notice on October 2 and gave no reason.
Why it matters for creators: If you built anything on Hume, export your voices, your settings and your scripts before the cutoff. And for everyone else: this is why I keep my source recordings myself, not just inside a tool.
Source: Voiceover Herald, October 18, 2026, citing Hume's developer changelog
What changed at ElevenLabs
▶ In the video at 04:33Quick note first: I'm an independent ElevenLabs affiliate, so the ElevenLabs link in this section is an affiliate link. It doesn't change your price.
Changelog of October 26 and November 2, 2026
- Professional voice clones on Eleven v4 now work for every workspace (November 2). A setting called clone strength controls how closely a v4 professional clone follows its samples. I tested instant against professional clones on v4 in October: ElevenLabs v4 Voice Clone: Does Instant Finally Beat Professional?
- One-credit minimum: since October 26, any text-to-speech or speech-to-text request that would cost less than one credit is charged one full credit. If you send lots of tiny requests, check your usage.
- Library voices in voice agents: conversations that use voices from the Voice Library now carry a separate surcharge. ElevenLabs says it goes to the voice actors.
And since Black Friday is coming: it's on November 27. Last year, ElevenLabs and Suno started their deals on the Monday before, which this year is November 23. Right now ElevenLabs already runs an offer: the first month of Starter costs $1 instead of $6, and the first month of Creator $11 instead of $22. It ends on November 16 (as of November 11, 2026). Until the real Black Friday deals start, my Black Friday overview for AI tools tracks what's announced.
Sources: ElevenLabs changelog · ElevenLabs pricing page
Quick hits
▶ In the video at 05:48HeyGen Voice Turbo is free through the API until November 30
HeyGen blog, updated November 2, 2026
HeyGen's new Voice Turbo model is free through HeyGen's API until November 30. HeyGen's price list shows $4 per million characters for Turbo. HeyGen says it's number one on the Artificial Analysis text-to-speech leaderboard. That's their claim. I'd want to hear it in a blind test before I believe it. My own blind test of HeyGen's voice model against ElevenLabs: HeyGen vs ElevenLabs Voice Cloning.
Source: HeyGen blog: HeyGen Voice Turbo
Google adds voice remixing to Gemini text-to-speech
November 4, 2026
You take one of Google's prebuilt voices, or a voice you designed, and change it with plain instructions. The result is saved as a new voice for Gemini 3.8 Flash TTS and Flash-Lite TTS. That's a way to get a consistent channel voice without cloning a real person.
Source: Google AI Studio changelog, entry of November 4, 2026
ElevenMusic publishes straight to SoundCloud
November 6, 2026
ElevenLabs now lets you publish tracks from its music tool, ElevenMusic, straight to SoundCloud. SoundCloud Artist Pro members get three months of ElevenMusic free, then $9.99 a month. So AI songs now reach a big platform with one click, in the same week the Elvis lawsuit argues about whose voice is in them.
Source: ElevenLabs blog: ElevenMusic and SoundCloud, November 6, 2026
Three things to do this week
If you remember one thing from this week: a voice doesn't have to be cloned from you to cause trouble. It only has to sound like you.
- Know where every voice you publish comes from.
- If you're on Hume, export your voices, settings and scripts before November 13, 12:01 a.m. Eastern.
- Set up a code word with your team and your family.
Coming up
Dates worth knowing about, as of November 11, 2026. The current list is on the AI Voice Weekly page.
- Nov 13Hume shuts down its text-to-speech and EVI voice APIs at 12:01 a.m. Eastern. Hume says any data left in your account is permanently deleted afterwards.
- Nov 16The current ElevenLabs offer ends: the first month of Starter for $1 instead of $6, Creator for $11 instead of $22 (as of November 11).
- Around Nov 23First Black Friday deals. Last year, ElevenLabs and Suno started theirs on the Monday before Black Friday. My Black Friday overview for AI tools tracks what's announced.
- Nov 27Black Friday. Cyber Monday follows on November 30.
- Nov 30HeyGen Voice Turbo stops being free through HeyGen's API. HeyGen's price list shows $4 per million characters for Turbo.
Full transcript
From the English subtitles of the video, grouped by chapter. Timestamps open the video on YouTube.
00:00 Who owns a voice that sounds like you?
A podcaster with millions of listeners says he heard his own voice in ElevenLabs' voice library, under a different name. ElevenLabs says it isn't him. And this week, Elvis Presley's estate sued Suno, partly over AI voices. So who owns a voice that only sounds like you? That's today's main story. After that: a $129 million scam with fake CEO voices, a voice company that shuts down on Friday and deletes your data, and what changed at ElevenLabs, including a deal that's already running. And there's one number from a new study that changed how I think about voice clones. I'm Lukas. I test AI tools and especially AI voice tools on this channel, and every Thursday I go through what changed in AI voice and what it means if you create with it.
If that's useful to you, subscribe.
00:50 ElevenLabs' Archer voice and Chris Williamson
Let's start with the voice that sounds like someone else. I've cloned my own voice for this channel in ElevenLabs, HeyGen, Voicebox and others, so this one is personal for me. Chris Williamson hosts one of the biggest podcasts out there. In October he posted a video about a voice called Archer in ElevenLabs' voice library. His friends heard it and said: that's you. He wrote to ElevenLabs. The answer, as he reads it out in his own video: a voice actor contributed Archer. ElevenLabs says it's checking whether that voice actor did anything fraudulent, but has no reason to believe so. And they ran a similarity test between Archer and his voice.
Above 0.7 would mean a match. Archer scored 0.28. Since then, Archer has gone from the library. Williamson says he doesn't know why. Nobody has sued anyone. But the question stuck.
01:39 Elvis estate sues Suno, Greene v. Google
This week it got a lot bigger. On November 6th, Elvis Presley Enterprises, together with heirs and publishers of songwriters, sued Suno in federal court in California. The complaint says Suno lets people produce songs that sound like Elvis singing, and it names Suno's voice feature, the one that lets you make songs with a saved voice. For the likeness claims, they put the damages at, quote, believed to exceed $50 million. That's an accusation, not a ruling. Suno hasn't had its day in court yet. But here's a case that did move forward. David Greene, a former NPR host, says the default male voice in Google's NotebookLM is his. Google tried to get the case thrown out.
On October 23rd, a judge in California said no, and all four of his claims go on. That doesn't mean it's his voice. The court hasn't decided that. It means the question is serious enough to fight about.
02:32 A study: 41% and what it means for you
And now the number I mentioned. A study from November 3rd, called VIPBench, had 1,290 listeners compare voices of 100 public figures. In 41% of the answers, people said a clone and a real recording of the same person were two different people. So a clone can fail to sound like the person it was made from. And if ElevenLabs is right about Archer, a voice can also sound like someone it was never made from. A similarity score is one answer. What listeners hear is another. And in both lawsuits, what people recognize is a big part of the argument. What this means for you. If a voice you use sounds like someone famous, I wouldn't publish with it.
Here's what I do: I use my own clone or a voice I know the source of, and I keep a note of where each voice came from. And the library voice can disappear. So if your whole channel runs on one, definitely have a backup.
03:23 A $129M voice scam and Hume shutting down
Which brings us to the people who use voices for something much worse. Last week, prosecutors in Manhattan announced charges against three men. They're accused of posing as company CEOs and lawyers, using AI to imitate their voices and faces, and getting companies to wire about $129 million. The prosecutors call the technology super intelligence. I just call it a convincing phone call, to be honest. If you run a business, or you're the person people listen to, this is the cheapest protection there is: agree on a code word with your team and your family. And no money moves because of one call. Hang up, call back on a number you already know.
Next, a company that's leaving. Hume, the company behind the Octave voice model, is shutting down its text-to-speech and EVI voice APIs on November 13th, at 12:01 in the morning, Eastern time. Hume says any data in your account is permanently deleted after that. If you're watching this when it comes out, that's only hours away. If you build anything on Hume, export your voices, your settings and your scripts today. And for everyone else: this is why I keep my source recordings myself, not just inside a tool.
04:33 ElevenLabs: v4 clones, credits, $1 offer
Now to ElevenLabs. Quick note first: I'm an independent ElevenLabs affiliate, so the ElevenLabs link below is an affiliate link. It doesn't change your price, but it helps my channel. Three changes. Professional voice clones now work with ElevenLabs v4 for every account, and there's a new setting called clone strength. I tested instant against professional clones on v4 last month. That video is on screen now. You can see it here. Second: since October 26, any text-to-speech or speech-to-text request that would cost less than one credit is charged one full credit. If you send lots of tiny requests, check your usage.
Third: if you build voice agents with voices from the voice library, there's now a separate surcharge, and ElevenLabs says it goes to the voice actors. And since Black Friday is coming: it's on November 27th. Last year, ElevenLabs and Suno started their deals on Monday before, which this year is November 23rd. Right now, ElevenLabs already runs an offer: the first month of Starter costs $1 instead of $6, and the first month of Creator $11 instead of $22. It ends on November 16th. I'll cover the real Black Friday deals next week, and I know that there's something coming from ElevenLabs already next week, so definitely stay tuned.
05:48 Quick hits and what's coming
A few quick ones. HeyGen's new Voice Turbo model is free through the API until November 30. After that it's $4 per million characters. HeyGen says it's number one on a public text-to-speech leaderboard. That's their claim. I'd want to hear it in a blind test before I believe it. Google added voice remixing to Gemini text-to-speech. You take one of their voices and change it with plain instructions, and it's saved as a new voice. That's a way to get a consistent channel voice without cloning a real person. And ElevenLabs now lets you publish tracks from its music tools straight to SoundCloud. SoundCloud Artist Pro members get three months of ElevenLabs Music free, then $9.99 a month.
So if you remember one thing from this week: a voice doesn't have to be cloned from you to cause trouble. It only has to sound like you. Three things you can do this week. Know where every voice you publish comes from. If you're on Hume, export today, and set up a code word with your team and your family. Coming up: November 13th, Hume shuts down. November 16th, the ElevenLabs offer ends. Around November 23rd, the first Black Friday deals are coming in. November 30, HeyGen Turbo stops being free. All sources are down in the description. The next episode is out next Thursday. Would you publish with a voice that sounds like someone famous?
Let me know, and see you again next week.
Related
A new episode every Thursday
6 p.m. CET, 12 p.m. ET, on my YouTube channel. And tell me in the comments: would you publish with a voice that sounds like someone famous?
Subscribe on YouTube