On Sunday 2 August, a suite of consumer-facing provisions under the EU Artificial Intelligence Act became enforceable, introducing mandatory transparency measures, compliance obligations, and regulatory powers that directly affect how AI-generated music is labelled and how AI developers handle copyrighted material.
Transparency and labelling
One of the most visible changes is a requirement that AI-generated or modified images, audio, video, and text must carry visible labels. Content must also feature embedded, machine-readable watermarks, enabling platforms and users to detect AI involvement. In recent weeks, new labelling systems for AI-generated music have emerged, alongside startups such as Genotone that offer ‘Proof of Human’ fingerprinting technology.
Comprehensive summaries of training data
Article 53 of the Act demands that providers of General-Purpose AI (GPAI) models trained on massive, broad datasets, such as OpenAI or Udio, maintain technical documentation and publish summaries of the data used. For music models operating in the EU, this could mean disclosing lists of songs included in training sets, though many AI music companies have so far resisted such transparency. The music industry, which has already won lawsuits against Suno in Germany and has long pressed for data disclosure, is expected to examine how this rule can be leveraged. European courts can now compel GPAI developers to produce the documentation mandated by the Act during legal discovery.
Opt-out policies
The law also requires GPAI developers to establish policies that detect, respect, and enforce rightsholder opt-outs. When a rightsholder places digital opt-out metadata on a song, AI companies must not train their models on that track for the EU market without a paid licence. Violations can result in significant fines and potential platform bans within the EU.
Compliance is already affecting music companies directly. Universal Music Group (UMG) noted in its 2025 annual financial filings that the EU AI Act could impose “significant obligations and costs related to monitoring and compliance.”
