Radio Listeners Can’t Spot AI Voices, but They Don’t Trust Them

Play a station promo for 1,326 radio listeners without telling them who recorded it, and most cannot say whether they heard a professional voice actor or an artificial intelligence system. Tell those same listeners afterward that the voice was AI-generated, and 20% turn against it, five times the 4% who sour on a human voice once its origin comes out. That reversal sits at the center of a new study from Crowd React Media, a radio audience research firm that pitted AI-generated voices against professional talent on U.S. commercial radio.

Rather than asking listeners what they think about AI in the abstract, the firm played them the real thing. Crowd React Media split 1,326 weekly radio listeners ages 18 to 45 into two groups for two identical spots, a station imaging promo and a Mother’s Day reminder, letting half hear a human voice actor and half hear the AI version without telling either group which one they had. Before anyone revealed the source, there were no statistically significant differences in listeners’ ability to say which was which. “Listeners can’t tell the difference until you tell them,” writes Katie Miller, founder of Crowd React Media and Vice President of Strategy at Harker Bos Group, in the study’s analysis.

1,326 Listeners Couldn’t Spot the AI Voice

Plenty of surveys ask consumers to weigh in on AI without ever letting them hear it next to a human performance. This one did, twice, and the two clips moved almost in lockstep across nearly every category researchers measured.

Near-Identical Scores Across the Board

Across both clips, AI and human voices scored within a point of each other on professionalism, authenticity, credibility, energy and likability, the study found. Overall appeal came out essentially tied, and in one of the two clips, the AI version actually edged the human read by a single percentage point.

The Only Place the Human Voice Won

Humor was the exception. For a promotional script built around a joke, 33% of listeners described the human delivery as funny, compared with 26% who said the same about the AI version, a seven-point gap that stood as the only statistically significant difference recorded in either clip. “The human voice was perceived as meaningfully funnier,” Miller says. “Comedy timing, inflection, and the organic quality of a human reading a punchline are things AI hasn’t fully closed the gap on yet, at least not in the ears of radio listeners.”

Favorability Flips the Moment AI Is Named

The biggest swing in the whole study showed up after the reveal, not during the blind portion. Among listeners told they had heard a human voice actor, 48% said they viewed the clip more favorably afterward, and only 4% viewed it less favorably. Among listeners told they had heard an AI-generated voice, 20% said their opinion became less favorable, versus 25% whose opinion improved. “The performance was the same,” Miller writes. “The perception shifted dramatically the moment people knew the source.”

Zoom out from the two test clips and the caution holds. Overall, 44% of respondents said they have a positive opinion of AI-generated voices in advertising and media, versus 26% who view them negatively and 30% who are neutral. Ask specifically about radio stations using AI voices, though, and the mood turns more guarded: one-third said they would feel less favorably toward a station that uses AI voices, 21% said they would feel more favorably, and nearly half said it would make no difference either way.

Some of that wariness has nothing to do with radio at all. Voice phishing attacks surged by 442% in 2025 due to AI-driven techniques, and callers arrive primed to distrust an unfamiliar digital voice long before a station ever plays one. Phone makers are racing to reassure them: Apple’s scam-coaching features built into iOS 27 now try to flag suspicious calls before money changes hands.

Where Are AI Hosts Already on the Air?

The technology is not hypothetical. Stations and networks in several countries have already put AI-generated voices to work, sometimes without listeners noticing until after the fact.

• CADA, Australia – introduced an AI-generated host built with ElevenLabs voice cloning technology and modeled after a real employee, who hosted a four-hour daily show for six months without listeners realizing she wasn’t human.
• Sveriges Radio, Sweden – cloned a presenter’s voice from only seven hours of content, an AI tool that can now read the name of any podcast or show and sound just like the real presenter.
• Futuri Media, United States – the Colorado-based software company unveiled RadioGPT, billed as the world’s first radio host entirely powered by AI.
• Prisa Radio, Spain – rolled out an AI personality named Victoria that engages listeners through Alexa-enabled devices, built as the network’s brand voice for football, across a network that pulls in 4 million listeners a day across 200 local stations.

The appeal for programmers is efficiency as much as novelty. A host who needs dozens of promotional spots recorded for upcoming events and contests would otherwise have to block out significant studio time, potentially interfering with show preparation. Producers can generate those spots instead using the host’s cloned voice, keeping the tone consistent while freeing the presenter to focus on live airtime.

None of this is happening in a vacuum for the people whose careers depend on their voices. Voice actors and dubbing artists have been organizing against uncompensated AI voice cloning, with more than 100 tracked movements by creative workers in about 25 countries, including Turkey, Argentina, Chile, India and South Korea, according to a Rest of World investigation into Hollywood’s AI dubbing push. Miller’s survey data suggests radio listeners draw a similar line. “They’re opposed to feeling deceived, and they draw a distinction between production tools and on-air personality,” she says.

Futuri Found the Same Blind Spot Two Years Earlier

Crowd React Media is not the first firm to catch listeners failing this test. Colorado-based Futuri Media reached a similar conclusion in its own research on AI voice perception in radio, surveying nearly 5,200 radio and TV news consumers across the U.S. in partnership with CMG Custom Research. That study found AI voices were often indistinguishable from human ones, with a 60% chance of misidentification, meaning listeners identified the AI voice as human and the human voice as AI more often than not.

“They’ll happily engage with AI, but you should be open about how AI is being used to create your content,” said Daniel Anstandig, chief executive of Futuri Media.

The broader numbers around AI voice adoption keep climbing alongside the studies:

• $4.06 billion – the projected size of the global AI voice cloning market this year, on pace to reach $9.56 billion by 2030.
• 1 in 5 – radio listeners who believe they’ve already encountered AI-generated content on the air.
• 0.9% – radio’s year-over-year audience decline, the smallest of any traditional format tracked by Borrell Associates.

Miller Calls Disclosure the Deciding Factor

Crowd React Media’s own broader research points the same direction. In a separate State of Media 2026 report, based on a survey of 1,094 U.S. adults conducted in March and April, the firm found that nearly eight in 10 adults expressed some level of distrust toward AI-generated content, while 42% said they had substituted AI tools for human-produced media in certain situations.

Other industries are already building formal scaffolding for that kind of transparency. Cognizant, the IT services and consulting firm, recently rolled out real-time AI governance tooling built to help enterprises document how their AI systems are used. Radio has nothing comparable that is standard across the industry.

Disclosure is the real variable. The performance gap is negligible. The perception gap after disclosure is substantial.

Miller wrote in the study’s analysis.

The debate stopped being theoretical for broadcasters this spring. NAB Show 2026, the National Association of Broadcasters’ annual trade event, put AI adoption at the center of its programming, treating it as standard operating infrastructure rather than an experiment.

Frequently Asked Questions

Does Any Law Require Radio Stations to Disclose AI Voice Use?

No radio-specific federal disclosure mandate turned up in current reporting, but the legal landscape around AI voice misuse broadly is moving fast. More than 20 U.S. states have passed or proposed laws addressing deepfake content and impersonation, and the European Union’s AI Act classifies deepfake misuse as a high-risk category requiring transparency.

Will AI Replace Human Voice Actors?

Not entirely, but the pressure on entry-level work is real. Experts have projected a 30% to 50% reduction in voice-acting jobs over the next decade because of AI adoption in the industry.

Can Voice Actors Earn Money by Licensing Their Voices to AI Companies?

Some already do, on their own terms. Voice AI jobs pay up to 85 times as much as traditional voice-over work, according to Julianna Jones, director of talent success at the voice marketplace Voices.

How Much Cheaper Is an AI Voice Than a Professional Voice Actor?

Substantially, at least in adjacent production formats. Research from IDC found that AI-generated voice and avatar production can cut content costs by 70% to 90% compared with traditional human-hosted production, a gap that helps explain why broadcasters are experimenting at all.

Is Radio Holding Up Better Than Other Media in the AI Era?

Better than most. Eight of the nine media platforms Crowd React Media tracks saw declines in the share of users engaging three to five days per week, while radio’s usage held essentially flat in the firm’s State of Media 2026 report.