Deezer tightens rules as AI-generated music floods streaming services
Deezer reports policy shifts and labeling to curb fraud and protect rights as AI-generated music uploads surge, affecting streaming patterns and payouts.
Deezer outlines its approach to AI-generated music
Deezer has begun labeling and filtering completely AI-generated music to address rising uploads and suspected abuse of streaming systems. The company excludes such tracks from algorithmic recommendations and curated playlists while still making them available to users who search for them. Deezer’s chief, Alexis Lanternier, described the policy as a balance between offering a comprehensive catalogue and preventing manipulation and unpaid streams.
Scale of uploads and who is driving growth
Deezer told staff and partners that uploads of entirely AI-generated songs have climbed dramatically, with the company reporting roughly 75,000 such tracks being uploaded daily as of late April. The platform attributes about 80 percent of those fully synthetic uploads to one creator tool, Suno, while other services and a handful of newer platforms also contribute. Deezer noted that some earlier sources have fallen off, but new entrants and licensing-focused tools continue to expand output.
Streaming patterns show little organic demand
Despite the torrent of uploads, genuine listener interest in AI-generated music remains small according to Deezer’s measurements. The service estimates that only about 0.5 percent of total streams come from users intentionally choosing AI-generated songs, and a larger share of plays often reflects suspicious behavior. Deezer says that in many weeks patterns of abuse account for as much as 85 percent of streams tied to synthetic tracks, although the figure can dip to around 60 percent when tactics shift.
Fraud, detection and transparency measures
A key concern for Deezer is mass uploads used to game playlists and royalties, sometimes involving hijacked accounts or bots. The company has developed detection tools that identify audio fingerprints typical of generative models and applies a visible label where a track appears to be fully AI-generated. Deezer reports a very low error rate for these labels and says wrongful markings occur in roughly 0.01 percent of cases, with erroneous flags overturned when rights holders provide proof.
Licensing, rights and industry tensions
Rights and licensing remain central to the debate over whether and how synthetic songs should be compensated. Some generative models face lawsuits alleging that training on copyrighted material was improper, while a few tools have struck licensing deals with major labels and publishers. Deezer’s stance is to limit the algorithmic amplification of tracks created with unlicensed tools but to treat demonstrably authentic listener plays the same for payouts, arguing that an industry-wide consensus on licensing standards is needed to settle broader questions.
Hybrid tracks and cases that complicate policy
Not all new releases are entirely synthetic, and Deezer is seeing many hybrid cases where human songwriters supply lyrics or production touches while other elements are machine-generated. The platform checks only the audio output, so it labels tracks that appear fully produced by a tool even when some creative roles are claimed by humans. Deezer says transparency about generated status is the priority, because listeners cannot experience a synthetic performer live and royalty splits should reflect who contributed to the final recording.
Product experimentation and commercial prospects
Beyond content moderation, streaming services are weighing ways to use generative AI for fan engagement, such as remix tools that let listeners rework songs. Deezer is in talks with rights owners to explore product features along those lines, while cautioning that consumer demand and monetization remain unproven. The company contrasts the limited uptake of playlist generation by text prompts with the potentially more disruptive applications of AI that create or reshape music itself.
Deezer’s broader financial position gives context to its choices, with the service reporting stronger direct subscriber growth in recent quarters and, for the first time in 2025, returning a profit on an annual basis. Executives argue that measured policies on AI content preserve the platform’s role as a comprehensive music service while trying to prevent algorithmic gaming and ensure rightful compensation. The company says it will continue refining detection, licensing engagement and product experiments as the industry develops shared rules for synthetic music.
