This is an edition of The Atlantic Daily, a newsletter that guides you through the biggest stories of the day, helps you discover new ideas, and recommends the best in culture. Sign up for it here.Late last month, a swarm of songs with near-identical names, lyrics, and melodies started to go viral on streaming platforms across the world. These tracks were not exactly the same—some have a little more guitar than others, some are more dance-oriented—but they’re all named something close to “Angel Above Me” or “Run Run River,” after the song’s first line. They’ve accrued millions of streams on Spotify and TikTok, and versions have hit No. 1 on iTunes in Germany and Austria.Many of them appear to have been generated by AI. It turns out that they’re based on a human-made song, “Angels Above Me,” which was released in 2019 by the reggae band Stick Figure. That track has enjoyed a streaming bump in recent weeks too—but many of the people listening to the new remixes may not even know about the original, because the song’s actual co-writers aren’t always credited. AI music has gone viral before and charted before, but song generators are now good and fast enough that they can flood the zone, creating tracks that slip past the safeguards of major streaming platforms and distributors. Spam-filtering systems can do only so much to stem the flow; according to data from the analytics firm Luminate, 106,000 songs (both AI-generated and not) were uploaded to streamers and other platforms every day in 2025.Musicians copy one another for legitimate reasons all the time. Parody law, as the comedy albums of Weird Al Yankovic remind us, is expansive; you can often get away with ripping off melodies and lyrics as long as it’s clear that you’re mocking them. Covering another artist’s song is also legal, as long as you get the right license before you do so. The same goes for sampling—knitting together different musical clips to create something new—and interpolating melodies. When Ariana Grande repurposed a tune from The Sound of Music on her song “7 Rings,” she ceded 90 percent of songwriting royalties to Rodgers and Hammerstein. The process of vetting your song in some way isn’t always simple or cheap, but it ensures that the original artists are getting paid as appropriate.AI remixes exist in a legal gray area. It took me all of 30 seconds to generate an AI clone of Kendrick Lamar’s voice for a “Not Like Us”–style diss track against the color blue. And it was just as easy to spin up an AI version of an Elliott Smith song featuring the exact lyrics of the original. Although I chose not to upload these to any streaming platforms for obvious legal and ethical reasons, I almost certainly could have: DIY distribution programs will push any song to Spotify, YouTube, Apple Music, or others for a small fee, and not all of them are equipped to vet each upload for potential copyright violations. That makes it all the more important for streamers to ensure that AI-generated music is both legal and properly labeled, and that a portion of the money flows to the original creators.The problem isn’t AI music itself. Many of the songs that AI generates could be legal, if they are distinct enough from human musicians’ published work to avoid copyright trip wires. And some artists, such as Timbaland, Kanye West, and Diplo, are now openly using it as part of their own creative processes. Few data are available about the total number of purely AI-generated songs on streaming platforms, but it’s clearly enough to spook creators. Takedown requests through the Digital Millennium Copyright Act—one of the main tools for fighting intellectual-property infringement—offer only a piecemeal remedy, striking songs one by one rather than banning unauthorized material wholesale.A representative for Suno, a leading AI-music-generation platform, told me that the company uses filtering technology to try to prevent unauthorized uses of artists’ preexisting songs, but she also said that the company didn’t know whether the unauthorized Stick Figure remixes had been created with their tools. A spokesperson for Spotify told me that “in the past year alone,” the company has “removed over 75 million spammy tracks from the platform” and “introduced a suite of new policies” regarding AI. She also wrote that “for any manipulated streams on Spotify, we remove those streams from play counts and withhold royalties.”Companies take different approaches to the problem of labeling AI-generated content, but in recent months, a consensus has emerged that it might just be easier to verify human-generated content instead. Adam Mosseri, the head of Instagram (another platform that’s rife with AI-generated material), wrote in December that as AI gets better at imitating reality, “it will be more practical to fingerprint real media than fake media.” OpenAI CEO Sam Altman is backing a start-up that scans people’s eyeballs in order to provide “proof” of humanness. And a few weeks ago, Spotify began rolling out a verification-badge system for artists who meet their “authenticity” criteria. For now, the system largely excludes purely-AI artists—although “artists who use AI tools responsibly” are eligible to be verified.These companies will need to fight the incoming wave of AI slop even as they lean into AI as a product. Last Friday, Universal Music Group announced a licensing deal with TikTok that promises to expand protections against AI music. But a day before, UMG had announced a partnership with Spotify that allows users to create AI-assisted remixes of certain songs. Spotify’s co-CEO Alex Noström described it as “grounded in consent, credit, and compensation for the artists and songwriters that take part.”Stick Figure’s Spotify page is now verified, but whether this will actually redirect listeners toward its music is unclear. The fuzzy line between human- and AI-generated material poses a particular problem in the streaming industry, which has long encouraged passive listening habits. As the writer Liz Pelly lays out in her book, Mood Machine, one of the ways Spotify makes money is by encouraging users to listen to music at all hours of the day; hence the hundreds of “chill” playlists designed to soundtrack our lives without intruding into the foreground.By framing music as sonic wallpaper, Spotify and other streamers have effectively set the stage for today’s confusion. We’ve been trained not to think all that hard about what we’re listening to. A recent study from Luminate found that people are growing less interested in AI-generated music—but what if they can’t tell they’re listening to AI-generated music in the first place? Plus, when people hear songs on streaming platforms or in social-media videos, they’re not typically thinking about the nuances of copyright law and royalty payments (Swifties are the possible exception to this rule). When AI enables unauthorized remixes that streamers don’t catch, human musicians and writers lose out. It’s not some coming danger; it’s already happening. And listeners may not even notice.Related:Nobody cares if music is real anymore.AI is democratizing music. Unfortunately.