For the past year, Spotify has been quietly purging tens of thousands of podcasts that advertised illegal online pharmacies. A report released Thursday by Senator Maggie Hassan, ranking member of the Joint Economic Committee, faults the company for acting only after news outlets exposed the content and her office spent nearly a year pressing for answers.
None of what it removed was sent to law enforcement, the report says.
Spotify reportedly removed more than 57,000 podcast episodes and 3,000 shows, and took enforcement action against 3,500 accounts, all pushing links to illegal online pharmacies advertising opioids, benzodiazepines, and stimulants for sale without a prescription. Nevertheless, the report frames the cleanup as a moderation failure.
The report leans on one comparison in particular: Spotify acted against more than 3,500 accounts for drug content in 2025 but fewer than 100 the year before. The committee presents the jump as evidence the company moved only after it came under scrutiny. Spotify offered a different explanation: that its older counts are incomplete because, as it says in the report, it changed the way it tracks removals last year.
A handful of the offending podcasts did find an audience. Of the five that drew more than 100 plays, two together pulled around 13,000 streams and walked listeners through buying modafinil, a wakefulness drug, by sending bitcoin. Another, with 125 plays, linked to sites posing as pharmacy marketplaces for cancer and HIV medications. Those were the exceptions, but they pointed to working ways to pay and order.
The numbers are alarming, and the stakes are real, Hassan says: Counterfeit pills bought online are frequently cut with fentanyl, and teenagers are among the most exposed.
“In the age of AI, all online platforms need to deploy sophisticated efforts to continually identify and take down illegal content,” Hassan tells WIRED. “Failure to swiftly detect and remove dangerous content and also report it to law enforcement can lead to harrowing consequences—whether that’s a teenager who buys drugs online that could be laced with deadly fentanyl or a senior who falls for a scam that wipes out their retirement savings.”
Asked about its approach to AI podcasts, Spotify spokesperson Laura Batey says the company “has a long history of working with law enforcement when content violates the law.” She did not say whether Spotify makes proactive referrals to the Drug Enforcement Agency, or how often. Batey said Spotify is still looking into WIRED’s question about whether it tracks clicks on those links.
Spotify told the committee that its practice is to alert authorities only when it identifies a credible threat of serious harm: an imminent risk to someone's life or safety. The podcasts, which it had classified as a search-optimization scheme rather than evidence of actual drug sales, never met that bar, the company said.
While Spotify did not say whether it reports illegal drug activity to the DEA, the report says the company's competitors answer that question directly: Snap regularly makes proactive referrals to the agency, and Meta says it cooperates with law enforcement to combat drug sales. Spotify's position, according to the report, is that, as a licensed-content streaming service, its obligations differ from those of a social network.
At least one of the removed podcasts pointed somewhere law enforcement was already looking. A show the committee flagged in July 2025—listed under a string of nonsense characters and titled to advertise a “licensed online vendor”—linked to a site called Opioidstores.com. That domain was later seized by federal prosecutors in Brooklyn, working with the DEA, the FDA, and other agencies. Spotify removed the podcast but, by its own account, reported nothing.
Of the episodes Spotify removed, the company told the committee, 94 percent drew zero plays and 99 percent had fewer than 10. The shows were barely heard, because reaching Spotify's audience was never the goal, according to the company, which says the actual payload was links buried in episode descriptions and cover art—an effort at exploiting Spotify's standing with search engines to push illicit pharmacy and scam sites up Google’s rankings.
Play counts, though, only measure whether someone listened to the audio, and by Spotify's own account, the audio was never the point. What the operators wanted was for listeners to click the links tucked into episode descriptions and cover art. And Spotify doesn't track those clicks. The company told the committee it monitors link activity only for ads it was paid to run, not for links inside ordinary podcasts. Its numbers can show that almost no one pressed Play but cannot show how many people followed a link to a pharmacy or scam site.
The same fake drug series turned up well beyond Spotify. Committee staff found copies on iHeart, Amazon Music, and Podchaser, several stamped with nearly identical 2021 upload dates. That overlap reflects how podcasting works. Shows are published once, and the various apps all pull from that single source. Removing it from one app does nothing to the original or to the copies running everywhere else.
Amazon Music and Podchaser did not immediately respond to requests for comment; iHeartMedia could not be reached.
Spotify told the committee it has several systems for catching this content. The company keeps a list of drug names and street slang and uses software to spot when a banned user opens a new account. It also runs new and edited episodes through an AI filter before sending questionable ones to human reviewers.
The company pays an outside firm, LegitScript, to review podcasts, though only once every three months. Content that gets pulled is removed from search immediately, Spotify said. Anything still appearing in search results is there because it hasn't found it to break the rules.
Much of this content is now generated by AI. The report points to services that market creative studios for AI podcasting—synthetic hosts, cloned voices, and a method for publishing straight to Spotify. In one case the committee flagged, an AI-generated podcast posing as a real psychiatrist ran episodes walking through benzodiazepines like midazolam and estazolam, drugs the DEA warns are misused by teenagers.
Spotify has built defenses against AI spam, though so far for music rather than podcasts.
In September 2025, Spotify announced new AI protections and said it had removed 75 million spam tracks over the previous year. It told the committee those measures were specific to music and that it has no policy against AI-generated podcasts. The report says a Spotify representative told committee staff the company is not well-positioned to identify AI-made content. A rival, iHeartMedia, went further in November, pledging that the podcasts it publishes are “guaranteed human.”
The problem does not stop at Spotify or even podcast platforms more broadly. The company told the committee that the same search-manipulation spam has spread across the internet, including onto local, state, and federal government websites. And as AI drops the cost of producing it, the content keeps surfacing wherever a trusted domain can be borrowed.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。