PARIS — Qobuz has begun placing a visible label on recordings its internal system identifies as generated by artificial intelligence, adding a new layer of disclosure—and a new potential source of conflict—inside music streaming.
The independent streaming service says its human editors will continue excluding AI-identified recordings from editorial playlists and featured placements. Qobuz reports that 60% of streams involving AI-identified tracks trigger its anti-fraud systems and are excluded from royalty calculations.
The company has also removed approximately four million unplayed AI albums from search. AI-identified music reportedly accounts for only 0.38% of total listening on the platform.
Detection should begin a review—not engrave a verdict in stone
RNN analysis: An automated detector should never become the sole and final decision-maker when a label can affect discovery, reputation or royalty payments.
A detection result should trigger one more step before a public label or financial consequence is applied. The platform should notify the creator or rights holder, disclose the basis for the classification and request confirmation or supporting documentation. If the creator disagrees, the appeal and evidence-review process should begin there.
That protection matters because recordings may combine human performances, conventional production tools and AI-assisted elements. A detection system could also produce a false positive. Qobuz has not publicly explained its classification standard, treatment of hybrid recordings or the process available to an artist who disputes a label.
Artificial-intelligence creation and streaming fraud are not interchangeable. Fraud concerns manipulated plays and false royalty generation. A recording’s production method does not, by itself, establish fraudulent activity.
Source: Barrett Media, September 25, 2026.