AI is having a transformative impact on content and advertising, and the audio space is no different.
But the medium is particularly proud of its intimate relationship with its listeners and the personal connection between the host and the audience. As AI tools are increasingly used to translate and edit podcast content – as well as streamline and improve audience targeting in audio – podcasters are struggling to find a balance between protecting the nature of podcasts and evolving the space for the AI era.
“Advertisers are looking for results, and [if] using AI helps me make a sale faster, helps me inform a consumer faster, helps me optimize faster, because I have more data coming to me more efficiently – so that’s perfect. “[Before AI] you needed a huge budget to be able to do things in multiple languages, to be able to have multiple different images or backgrounds. All of this can be done so effectively today.
But integrating AI technology into podcasts poses transparency, ethics and regulatory compliance issues, executives said.
“People have such an expectation of personal connection that we are at real risk as an industry if we start to widely deploy AI in a way that actually impacts content or ads without recognition,” Matt Shapo, IAB director of digital audio and video, told Digiday.
The podcast industry is still figuring out AI transparency
As a tool, generative AI can be useful for replacing a mispronounced or mispronounced word, or for removing filler words from a podcast episode.
Sonoro Media also uses these tools to find key moments in a podcast episode for social clipping and marketing purposes, Camila Victoriano, co-founder and chief content officer of Sonoro Media, said on stage.
For these processes, AI tools mirror editing tools like Microsoft Word — a tool that people don’t need to give credit for, Shapo said.
“I don’t see why anyone should disclose that. You edit your podcast the same way you edit a story that you put online,” Shapo said.
Podcasters also use generative AI tools to translate shows into other languages.
Sonoro translated a Spanish-language horror podcast called “Relatos de la Noche” into English using AI and the podcast host’s own voice, complete with a Spanish accent.
There are no industry standards for disclosing the use of AI in podcasting – or otherwise. But Sonoro pointed out in the show’s content and in the podcast’s show notes that she used AI tools to translate the podcast.
“There was nothing to hide. We were very transparent from the start. We made it clear that it was an AI translation. [and mentioned] the software we use. In the content, you always credit the engineers, producers and writers. Why not credit these systems that help you achieve this? Victoriano said.
AI disclosure is a topic that professional bodies like the IAB are currently working on, but at the very least, podcasters should disclose the use of AI in their show notes, Shapo said. And whether content or advertisements are generated by AI to ensure these providers comply with regulations, he noted.
It is unclear how to isolate audio content from AI scraping
Publishers are increasingly taking steps to try to protect their content from removal and uncompensated use by AI tools, but it is not yet clear how they access podcasts.
“When the time comes, we absolutely want to participate in protecting our intellectual property,” Victoriano said. “I want to protect the creators that we work with and I really want to be on them to define what they’re comfortable with until they have more knowledge and more education.”
Sonoro CEO Joshua Weinstein told Digiday that Sonoro’s podcasts are available on RSS feeds, somewhat insulating their content from what happens to publishers. It’s “much harder” to scrape content from podcast feeds compared to content from a newsletter or website, he said.
“It’s harder to steal and manipulate audio. But it’s not impossible. It’s getting easier,” Weinstein said.
