Podcast Index: AI now makes almost 40% of new podcasts, and the audio industry is worried

The audio industry has a new problem: a flood of podcasts that neural networks assemble end to end and that get published with almost no human involvement. According to Podcast Index, roughly 39% of new shows added to the open catalog over the past nine days were likely created by AI.

Where the numbers come from

These aren't isolated experiments with synthetic voices. AI shows now make up a noticeable share of all new submissions. In one April snapshot covering 24 hours, Podcast Index flagged only 44.6% of new shows as "likely legitimate" and 45.7% as potentially AI-generated. The index classifies shows with its own detection tools, which means the industry has already reached the point where it has to use AI to track AI.

The catalog has also started rolling out infrastructure measures. Podcast Index now offers a separate problematic API that lists new feeds flagged as spam, phishing or low-effort AI. The idea is simple: hosting providers, apps and aggregators can filter out junk faster, before it clogs search and recommendations.

This isn't happening on a single closed platform. It's happening in the open part of the podcast ecosystem, where many different apps and directories pull in the same RSS feeds. If the noise takes hold here, it will spread further: into recommendations, into charts on niche services and into ad tools that respond to the sheer volume of new shows rather than their actual value to listeners.

Who's pumping out the content

One of the most visible players is Inception Point AI. According to industry reports, the company managed to release 325 new shows in a single day, nearly a fifth of all new podcasts that day. The catalog attributes more than 8 thousand active shows to it. Its wellness lineup alone added 20 programs in under 75 minutes, covering topics including mental health, addiction and vaccination.

"I'm an AI host, so I can talk about sensitive topics without judgment

or a hidden agenda."

The volume is only part of what's alarming; the business model is the other. Back in fall 2025, an Inception Point AI co-founder told Bloomberg that the company publishes about 3 thousand episodes a week and turns a profit if an episode gets just 20 listens. That changes the barrier to entry. Launching a "media network" no longer requires a studio, an editorial team or regular hosts. A text generator, speech synthesis and cheap distribution are enough.

Why the industry is nervous

AI in audio production no longer shocks anyone on its own: people already use it for editing, dubbing, translation and voice restoration. The trouble starts when the technology isn't there to amplify a creator but to stamp out shows endlessly with no meaningful editorial oversight. That's the wave the industry has started calling podslop.

• Real shows get harder to find among thousands of near-identical feeds
• Recommendation algorithms get noise instead of signal
• Open catalogs take on extra costs for storing, parsing and refreshing RSS feeds
• Health, finance and news topics become riskier because of factual errors and hallucinations

For independent creators, it's a double hit. First, new launches lose visibility, since they're easy to bury under a flood of auto-generated titles and descriptions. Second, listeners have a harder time telling whether a voice is backed by a person with experience and accountability, or just by a cheap content farm optimized for keywords.

There are already concrete quality concerns. In Spanish-language biography shows released by Inception Point AI, listeners found English fragments that looked like internal prompts or error messages from an AI tool. It's a telling marker of where the market stands: producing audio has become too easy, and neither creators nor platforms can keep up with checking the final product.

What it means

Podcasts are entering the same phase that search, SEO sites and stock media went through earlier: the volume of synthetic content is growing faster than the mechanisms for trust and filtering. For platforms, it's a signal to step up labeling and moderation. For independent creators, it's a reason to bet on a recognizable voice, real expertise and a format that's hard to mistake for an automated feed.