Media Morpheus

Data Reportby Media Morpheus Research

October 5, 2026

Would Listeners Accept an AI Version of a Podcast Host?

Edison Research asked weekly US podcast consumers how they would feel if their favourite host started using an AI-generated version of themselves. Two thirds picked a negative word, a third said “deceived”, and 18 per cent were indifferent. Apple already requires AI hosts and synthetic voices to be disclosed in every episode; YouTube does not require disclosure of a creator cloning their own voice. For a company thinking about an AI-voiced executive podcast, the survey and the rules point the same way.

PodcastsAIPodcast HostsDisclosureExecutive Visibility

Key Findings

  • Edison Research at SSRS added the question to its Edison Podcast Metrics survey of US weekly podcast consumers aged 13 and over in the first quarter of 2026 and published the results on 30 September. Respondents could choose more than one word.
  • The most common reactions were “deceived” (33 per cent), “upset” (33), “skeptical” (32) and “betrayed” (28). In all, 66 per cent chose at least one of those four.
  • Positive reactions were smaller: “excited” and “impressed” 14 per cent each, “optimistic” 12, “fascinated” 10. Edison puts the share choosing any positive word at 27 per cent. Eighteen per cent said they would be indifferent.
  • Apple Podcasts' content guidelines require creators who use AI to generate audio or video, “including synthetic voices, AI-generated hosts or on-screen personas, or AI-synthesized replicas of real individuals”, to disclose it prominently in the content and metadata of every episode and show.
  • YouTube requires disclosure of realistic AI-altered content but lists “cloning one's own voice to create voice overs or dubs” among uses that need none. Separately, its monetisation rules bar channels using AI personas that present as human experts on sensitive topics, naming “AI-generated podcast hosts offering financial guidance”.
  • The best-known attempt at AI podcasting at scale has shrunk: Podnews reported on 29 September that Inception Point AI's network now has 300 active shows it no longer tries to monetise, and that its chief executive described some of its publicity as “a PR strategy to get attention”.
  • Our reading: the people most likely to object are a show's regular listeners, which is the audience an executive podcast exists to build. That conclusion is ours, not Edison's.

What Edison asked

Edison Research, now part of the survey firm SSRS, runs Edison Podcast Metrics, a continuous survey of the US podcast audience used by publishers and advertisers. In the first quarter of 2026 it added custom questions about AI. One asked weekly podcast consumers how they would feel “if your favorite podcast host started using an AI-generated version instead of themselves to host their podcast”. It offered a list of reactions and allowed more than one answer. Edison published the results in its weekly insights on 30 September 2026 under the title “Hands Off My Host, AI”.

Reaction if a favourite host used an AI-generated version of themselves

Per cent of US weekly podcast consumers aged 13+, Q1 2026; more than one answer allowed

Deceived
33%
Upset
33%
Skeptical
32%
Betrayed
28%
Indifferent
18%
Excited
14%
Impressed
14%
Optimistic
12%
Fascinated
10%

Source: Edison Research at SSRS, Edison Podcast Metrics U.S., Q1 2026, published 30 September 2026.

66 per cent chose at least one of deceived, upset, skeptical or betrayed. Edison's text gives 27 per cent choosing any positive emotion; its text does not mention “excited”, which appears in its chart at 14 per cent.

The negative answers cluster tightly. “Deceived” and “betrayed” are words about trust rather than quality: they describe how listeners would feel about the host, not whether the AI version sounded good. The case usually made for host-read advertising leans on the same relationship between listener and host; we looked at how much published evidence supports that case in whether host-read podcast ads work better than produced spots.

How far the numbers go

Three limits matter. First, the question is hypothetical and personal: it asks about a favourite host being replaced, not about a show that was AI-hosted from the start and chosen knowingly. A listener might object to the first and still listen to the second. Second, the sample is weekly podcast consumers, the most engaged part of the audience; occasional listeners may care less. Third, Edison's post does not give the sample size, field dates within the quarter or exact wording of the answer list, and we have not seen the questionnaire.

Within those limits, the finding is unambiguous in direction. Negative reactions outnumber positive ones by about five to two, and fewer than one in five weekly listeners say they would not care.

What the platforms require

Apple Podcasts. Apple's content guidelines for creators include a rule headed “Artificial Intelligence (AI) Generated Content”. It says creators using AI to generate audio or video content, “including synthetic voices, AI-generated hosts or on-screen personas, or AI-synthesized replicas of real individuals”, must “prominently disclose this to audiences in the content and metadata for each episode and show”. The next rule bars using AI to mislead, including fabricating news or manipulating clips. An executive's voice clone reading a weekly episode would fall within the first rule on its plain wording: it is a synthetic voice and an AI replica of a real individual. YouTube. YouTube's rule is narrower. It requires creators to disclose AI use when it makes realistic content that could mislead, such as making a real person appear to say something they did not. But its help page lists “cloning one's own voice to create voice overs or dubs” among the examples that do not need disclosure. On YouTube, then, an executive narrating with a clone of their own voice need not label it, while the same episode in Apple Podcasts must be labelled.

YouTube's monetisation policies add a separate restriction. Channels that use AI-generated personas presenting as human experts on sensitive topics such as health, law, finance or politics are not allowed to monetise, and the examples include “AI-generated podcast hosts offering financial guidance, investment tips, or wealth management advice”. A company in financial services considering an AI host would meet that rule directly.

Whether advertising law also requires disclosure depends on the market and the content. We covered how the EU's AI Act treats AI-generated material in does the EU AI Act require labelling AI-written content.

What happened to AI podcasting at scale

The most prominent bet on AI-hosted podcasts was Inception Point AI, which attracted attention for producing thousands of AI-made shows. On 29 September 2026 Podnews reported, from an interview on the trade podcast The Lead Balloon, that its Quiet Please network now has 300 active shows that it is no longer trying to monetise, that the company has rebranded as Liquid Studios and describes itself as transforming existing content rather than generating from scratch, and that chief executive Jeanine Wright said some of its publicity had been “a PR strategy to get attention”. Wright told Podnews that “the catalog itself was never a stunt”. We have not heard the interview and report it as Podnews did.

That is one company, and it does not show that AI-hosted shows cannot find audiences. It does show that volume alone did not produce a business, which is consistent with Edison's finding that listeners value the person behind the voice.

What this means for a company podcast or an executive's voice

What follows is our reading, not Edison's or the platforms'.

The audience that objects is the one you want. A company podcast, or an executive's regular guest appearances, exist to build familiarity with a real person. Edison's respondents are exactly the regular listeners that familiarity depends on, and their strongest reactions are about deception. If AI is used, label it, and expect to have to. On Apple Podcasts disclosure of a synthetic host or voice is mandatory in every episode's content and metadata. Labelling on Apple and not on YouTube, which YouTube permits for an own-voice clone, would put two different accounts of the same episode in public. Production help is a different question. Neither Edison's question nor Apple's rule is about AI used to edit audio, write show notes or draft an outline. YouTube explicitly treats script and outline assistance as not needing disclosure. The line both draw is at the voice and the person. Guest appearances stay human. A podcast guest spot is a conversation with a host. None of the platform rules contemplates sending an AI version of a guest, and the Edison findings suggest audiences would react to it as they would to an AI host. Why the guest's own name and voice matter for later discovery is covered in whether saying your name and company on a podcast helps AI understand who you are.

Methodology and limitations

This guide is based on Edison Research at SSRS's weekly insight “Hands Off My Host, AI” of 30 September 2026, including the chart image published with it, read on 5 October 2026; Apple Podcasts for Creators' content and subscription guidelines, rules 1.11 and 1.12; YouTube Help's page on disclosing use of generative AI content; and YouTube's channel monetisation policies, section on AI personas related to sensitive topics, all read on 5 October 2026. The Inception Point AI material is secondhand from Podnews's report of 29 September 2026; we did not hear The Lead Balloon interview it describes. Four limitations apply. Edison's post does not publish the sample size, field dates or full answer list, and the figures are self-reported reactions to a hypothetical question. The survey covers US weekly podcast consumers aged 13 and over only. Edison's text and chart differ slightly: the text omits “excited” (14 per cent in the chart), and we have taken individual figures from the chart and the positive total from the text. And platform rules change; those quoted were in force when read.

Conclusion

Edison's question is narrow, but its answer is clear: asked about an AI version of their favourite host, two thirds of weekly US podcast consumers reached for words like deceived, upset and betrayed, and fewer than one in five said they would not mind. Apple already treats a synthetic host or voice as something that must be disclosed in every episode, and YouTube refuses monetisation to AI personas giving financial and other sensitive advice, though it lets creators clone their own voice without a label. For a company, the practical line runs between AI that helps produce a show and AI that stands in for the person. The first is routine. The second must be labelled on Apple and, on this evidence, risks the trust the podcast was meant to build.

Frequently Asked Questions

Do podcast listeners accept AI hosts?

Most regular listeners say they would not welcome one replacing a favourite host. In Edison Research's Q1 2026 survey of US weekly podcast consumers, 66 per cent chose at least one negative reaction, led by deceived and upset at 33 per cent each, while 27 per cent chose a positive one and 18 per cent said they would be indifferent.

Do I have to disclose an AI voice on Apple Podcasts?

Yes. Apple's content guidelines require creators using AI to generate audio or video, including synthetic voices, AI-generated hosts and AI replicas of real individuals, to disclose this prominently in the content and metadata of each episode and show.

Does YouTube require disclosure if I clone my own voice?

Not on its own. YouTube's help page lists cloning one's own voice to create voice-overs or dubs as an example that does not need disclosure. Realistic AI content that could mislead, such as making a real person appear to say something they did not, does.

Can an AI-hosted podcast make money on YouTube?

It depends on the content. YouTube's monetisation policies bar channels that use AI personas presenting as human experts on sensitive topics, and give AI-generated podcast hosts offering financial or investment advice as an example.

Is using AI to edit a podcast the same as an AI host?

No. Edison's question and Apple's rule concern AI generating the voice or the host. Editing, transcripts, show notes and outlines are different uses; YouTube explicitly lists script and outline help as not needing disclosure.

What happened to Inception Point AI's podcasts?

According to Podnews on 29 September 2026, its Quiet Please network now has 300 active shows it no longer tries to monetise, and the company has rebranded as Liquid Studios. Its chief executive said some publicity had been a PR strategy, while telling Podnews the catalogue itself was never a stunt.

Sources

  1. survey research Edison Research at SSRS, Hands Off My Host, AI, Edison Weekly Insights, 30 September 2026, with chart from Edison Podcast Metrics U.S., Q1 2026. Read 5 October 2026.
  2. platform documentation Apple Podcasts for Creators, Content and subscription guidelines, rules 1.11 (AI generated content) and 1.12 (misleading use of AI). Read 5 October 2026.
  3. platform documentation YouTube Help, Disclosing use of GenAI content, and YouTube channel monetization policies, section on AI personas related to sensitive topics. Read 5 October 2026.
  4. trade press Podnews, Was Inception Point AI just a PR stunt?, 29 September 2026. Secondhand account of an interview on The Lead Balloon. Read 5 October 2026.