Should you tell your audience you use AI on your podcast?
by Danny Abdeljabbar, founder of manekuThe short answer
Yes. Say it clearly on the episode where the AI first appears, and keep a line about it in your show notes. After that, the rule I follow is simple: never let an AI pass as a person. You don't need to announce it every time it speaks, as long as nobody could reasonably mistake it for a human. If you publish on YouTube, check its rules on labelling realistic synthetic content, which can cover AI voices.
I've co-hosted Bro History with my friend Henry for ten years and close to 350 episodes. When we started bringing an AI co-host onto the show, I worried about this question more than any technical one. Here's what we did and what happened.
What we did
For the first couple of episodes with the AI, we explained plainly on air that we had a robot on the show and what it was. We also put it in the show notes, partly because I'm happy to promote the thing I built, and partly because I think disclosure is simply the right default.
After those first episodes, we stopped announcing it. That was a deliberate choice, not laziness. Stopping the show every time the AI is about to speak to say "heads up, a robot is about to talk" would wreck the flow of a conversation. More importantly, nobody could mistake it for a person. It has its own name, its voice is clearly synthetic, and on video it's a named tile with a voice visualizer.
What our audience did
Honestly, almost nothing happened.
We saw no drop in listeners or downloads on our long-form episodes. Not a single comment on those episodes mentioned the bot at all. A couple of random people commented on short clips to say they didn't like AI, which is about the same rate of negative comments we get on everything else.
I'll be upfront about the limits of that. Bro History is a small show, a few thousand listeners per episode. I don't know how a much bigger audience would react. And part of me was a little disappointed that nobody said "hey, cool bot."
Where I draw the line
For me, the question isn't really "did you announce it." It's "could anyone be fooled." I think that's the right line for a few reasons.
- Impersonation is the actual harm. An AI that sounds like a real, specific person, or that's presented as a human guest, misleads your audience. An AI that's clearly an AI doesn't.
- Once, clearly, beats constantly, vaguely. A proper introduction and a line in the show notes tells listeners more than a muttered disclaimer every few minutes.
- Your audience's trust is the whole asset. If people later found out that a "guest" was synthetic, that's a much bigger problem than a one-line disclosure ever could be.
What the platforms say
Platform rules are moving, so check the current versions yourself. The one most podcasters run into is YouTube. It asks creators to disclose realistic content that's been altered or made synthetically, including things like a generated voice that viewers could mistake for a real person. Clearly unrealistic or obviously synthetic content generally doesn't need the label. You set it in YouTube Studio when you upload. If you're unsure whether your AI co-host counts, ticking the box costs you nothing.
A simple way to do it
If you want a template, here's roughly what I'd do:
- On air, the first time: "We've got an AI co-host on the show today. It's called [name]. It listens to the conversation and only speaks when I ask it to."
- In the show notes, every episode: "This episode features [name], an AI co-host."
- With guests, before recording: tell them there's an AI in the room and what it does.
- On video: keep the co-host's name visible on its tile.
That's it. Be clear, be consistent, and never let it pretend to be someone it isn't. For the full consent side of this, including what to tell guests, see our AI disclosure and consent guide.
Questions people ask
Do I have to disclose AI use on a podcast?
Rules depend on the platform and where you live, and they're changing. YouTube asks creators to label realistic altered or synthetic content. Beyond any rules, I think you should disclose because your audience's trust is the whole asset.
Do I need to announce the AI every time it speaks?
I don't think so, as long as it's obviously not a human. Constant announcements kill the flow of a show. Introduce it clearly once, keep it in your show notes, and make sure its name and voice make it unmistakable.
Will listeners leave if I add an AI co-host?
In my case, no. We saw no drop in listeners or downloads after adding one to a ten year old show. That's one small show, so your results may differ, but it wasn't the backlash I'd braced for.
Should I tell guests there's an AI in the room?
Yes, before you record. It's basic courtesy, and some guests will want to know how their words are being used. maneku has a disclosure and consent guide with wording you can borrow.
Is it okay to clone a real person's voice for my show?
Not without their clear permission. That's impersonation, and it's where disclosure stops being a courtesy and becomes a real ethical and legal problem.
Danny Abdeljabbar
I built maneku. I also make all the artwork and marketing for Dry Quest, an app I developed with my wife, and I'm usually building a smaller app or two on the side. I co-host the Bro History podcast, and by day I work in revenue operations. Before all of this, I was an artist across a lot of mediums, including music and audio.
Want a co-host in your next recording?
maneku is in open beta on macOS. It joins Descript, Riverside and Restream as a real participant.
Get early access