Earlier this year, an inflection point arrived quietly. By June 2026, the French streaming service Deezer was processing an average of 90,000 AI-generated tracks every single day. That volume represented more than half of all daily uploads to the platform. What began as a trickle in January 2025, when synthetic music accounted for merely 10 percent of incoming uploads at roughly 10,000 tracks per day, has become a torrent that is actively reshaping the economics of digital music.
The Upload Curve
Deezer has been monitoring this wave since early 2025, and the trajectory tells a clear story. Nine months after that initial 10 percent baseline, the share had nearly tripled. By September 2025, AI-generated content represented 28 percent of daily uploads, totaling approximately 30,000 tracks. Seven months later, in April 2026, the figure hit 44 percent, translating to about 75,000 new synthetic tracks each day. The climb did not level off there. It accelerated into June 2026, when the daily average reached 90,000 AI-generated uploads and finally pushed the total past the 50 percent threshold.
These are not just abstract statistics for platform engineers. Each synthetic track enters the same royalty pool as music written and performed by humans. When the denominator explodes, the per-stream value drops for everyone else. This is the mechanics of payment dilution: more tracks competing for the same listener hours and the same fixed pot of subscription revenue. For independent musicians who rely on streaming income, the arrival of 90,000 non-human competitors per day is not a theoretical concern. It is a direct threat to their paycheck.
Drawing a Harder Line
Deezer CEO Alexis Lanternier has stressed that the company is taking proactive steps to protect songwriters and artists. The response is not a blanket ban, but a targeted purge based on two simultaneous criteria. First, the platform is removing AI-generated tracks that have logged zero streams over the previous six months. Second, it is taking down synthetic content that has been flagged as part of fraudulent streaming operations, schemes where bots or click farms artificially inflate play counts to harvest royalty payments.
This dual requirement matters. An inactive track does not necessarily mean a fraudulent one. By pairing the six-month dormancy rule with confirmed fraud signals, Deezer attempts to avoid punishing legitimate experimentation while clearing out the digital debris that clutters search results and drains resources. The policy acknowledges a practical reality: not all AI music is malicious, but much of it is disposable, and the platform cannot afford to host an ever-expanding graveyard of unlistened files.
The industry remains divided on how to handle this. Bandcamp has chosen an outright prohibition on AI-generated content. Apple Music, meanwhile, has leaned on voluntary tagging systems that rely on uploaders to self-identify synthetic material. Deezer sits somewhere between these poles, accepting the technology’s presence but asserting aggressive moderation.
Naming the Tools
Behind the takedowns sits a detection stack that Deezer has been refining for over a year. The system does not merely flag suspicious audio patterns. It can identify tracks produced by specific controversial startups, namely Suno and Udio, both of which are currently facing high-profile copyright lawsuits over the training data used to build their models. That specificity is unusual. Instead of treating all synthetic audio as a monolithic blob, Deezer’s technology fingerprints the output of particular generators.
The company has not kept this capability entirely in-house. In an effort to raise the floor for the entire streaming ecosystem, Deezer has opened its detection technology to other platforms and recently released a tool designed to scan third-party playlists. The software can sift through public playlists on Apple Music and Spotify, flagging which entries appear to be AI-generated. Whether those larger rivals have actually adopted Deezer’s tools is still an open question. The release itself, though, raises the pressure. It forces a conversation about cross-platform transparency that Spotify and Apple have so far avoided.
Bagi pengembang dan spesialis keamanan siber, hal ini menandakan pasar yang terus berkembang. Jika platform berukuran menengah seperti Deezer dapat membangun pengklasifikasi yang dapat berfungsi untuk audio sintetis, permintaan akan semakin meningkat untuk watermarking terstandarisasi, verifikasi asal-usul, dan autentikasi konten secara real-time. Generasi berikutnya dari manajemen hak digital mungkin tidak akan terlihat seperti kunci DRM, melainkan lebih seperti audit rantai pasokan, yang memverifikasi dari mana sebuah berkas suara berasal sebelum sampai ke daftar putar pendengar.
Mengapa Platform Tidak Dapat Mengalihdayakan Masalah Ini
Pelajaran bagi ekosistem teknologi yang lebih luas adalah bahwa AI generatif tidak mendistribusikan biayanya secara merata. Para pencipta model pembuatan musik meraup keuntungan dari skala: biaya marginal yang rendah, replikasi tanpa batas, dan tanpa waktu studio. Platform hilir menanggung eksternalitasnya: biaya penyimpanan, penurunan kualitas pencarian, investigasi penipuan, dan kerja politik yang berat untuk menjelaskan kepada para seniman manusia mengapa porsi keuntungan mereka terus menyusut.
Keputusan Deezer untuk melacak, mengungkapkan, dan secara aktif membatasi kategori konten AI tertentu adalah sebuah pengakuan bahwa aturan lama konten