What actually changed on July 15?
TIDAL's detection system now flags any track it identifies as 100 percent AI-generated, and that track stops earning money, full stop. An "AI" badge appears next to it in the app, and it's locked out of both royalties and direct-to-fan sales. The rule itself has technically been running since June 29, but this week is when it became visible and enforced across the catalog. The line TIDAL draws is narrow on purpose: only tracks with zero human input on both vocals and instrumentation lose eligibility. A producer running a vocal chain through an AI stem separator, or building a track around an AI-generated loop they then chopped and arranged themselves, keeps getting paid like anyone else.
Why does this land harder on TIDAL than anywhere else?
TIDAL has quietly paid more per stream than every other major platform for years, somewhere around $0.012 to $0.015, three to five times what Spotify pays out. That made it, structurally, the best possible place to dump an AI-generated track: same distribution effort, a bigger per-stream check. There's a second reason AI tracks were unusually profitable there and everywhere else. A wholly machine-made track has no human author, and under both US and EU copyright law, work without human authorship isn't copyrightable. No songwriter, no publisher, no composition to split revenue with. So whoever uploaded it kept a bigger cut of every stream than a human artist releasing through the exact same platform ever could. TIDAL's ban doesn't just stop the fraud angle, it kills that structural advantage outright on its own platform.
Does cutting off TIDAL actually shrink the fraud problem?
Only on TIDAL. Deezer, which built the industry's first AI detection system back in early 2025, has since tagged more than 13.4 million AI-generated tracks and rolled out a second-generation detector this June that it claims hits 99.8 percent accuracy; it's now licensing that tech to rival platforms. But Deezer and Qobuz still only pull royalties when fraud is actually detected on a track, they don't ban wholly AI-made music outright the way TIDAL now does. Spotify and Apple Music are looser still: Spotify's revamped policy from September 2025 pulled more than 75 million tracks flagged as spam or AI-fraud over twelve months, and Apple Music only started AI transparency tags in March 2026, but both keep paying royalties on any AI track that isn't specifically flagged. Meanwhile IFPI puts the annual cost of streaming fraud, much of it AI-driven, at roughly $2 billion in lost artist royalties, and Deezer's own numbers show 44 percent of everything uploaded daily, about 75,000 tracks, is fully AI-made, with 85 percent of the plays those tracks generate coming from bots rather than listeners. TIDAL calls the policy a living document, meaning it expects to tighten it as the detection tech improves. For a working deep house or tech house producer, the honest read is: the platform that used to pay you best just closed the AI loophole on itself, but the pro-rata pool on every other service is still absorbing bot-driven AI volume exactly as before.
"AI is the ultimate enabler" of streaming fraud, IFPI's Chief Global Content Protection Officer Melissa Morgia told a WIPO enforcement panel, because it lets bad actors operate at scale while staying under the detection threshold on any single track.
