The initiative in brief

In May 2026 the Commission opened a call for evidence for a targeted legislative initiative – expected as a directive – to adjust the EU copyright framework for the competitiveness of the creative sectors and for innovation, in parallel to the Article 30 review of the Copyright in the Digital Single Market Directive (EU) 2019/790. A legislative proposal is provisionally scheduled for the first quarter of 2027. The initiative covers four areas: challenges faced by creators linked to artificial intelligence, including licensing between rightsholders and AI providers and AI-generated imitations of performers; online piracy of time-sensitive content including live events; the remuneration of music performers and producers for sound recordings of third-country nationals; and the use of protected content in scientific research, with a possible EU-wide secondary publication right. Building on this call, the Commission launched a targeted consultation on 29 September 2026. The context is well documented: the text and data mining opt-out of Article 4(3) has not delivered on its central promise – rightsholders who opted out struggle to enforce their rights, to license and to be remunerated – while the AI Act obliges providers of general-purpose AI models to comply with EU copyright law and to publish training-data summaries, and machine-readable opt-out protocols are being developed with the support of the EUIPO.

Five points GISAD recommends

1. An opt-out that nobody reads is not an exercise of rights. On paper, the TDM opt-out gives rightsholders control; in practice it is a leaf in the wind. Machine-readable, standardised opt-out protocols must be mandatory, not optional. And the training-data summaries under Article 53 of the AI Act must be verifiable: whoever verifies must be verifiable themselves. Logging at a neutral instance, examination outside the manipulable area.

2. Structural remuneration instead of gatekeeper licensing. Small rightsholders must not be left to individual negotiations with AI corporations – negotiations they lose. Collective licensing, collecting societies and micropayment infrastructures turn remuneration into structure. Participation applies to creators as well.

3. AI imitation of performers: structure instead of hindsight. AI-generated fakes must be recognised structurally, not litigated individually: AI detects AI fakes at the point of publication, and every publication carries a signed instance number – responsibility becomes assignable without building profiles. Concept fidelity arises through measurement, not through examples.

4. Live-event piracy without filters 2.0. Enforcement is right – but the answer must not be a revival of the upload-filter debate. Structure instead of filters: signed transmission paths and provenance evidence, while the audience remains WAN-anonymous. Filters recognise copies – structure recognises origin.

5. Research as an infrastructure of participation. Preserve the TDM research exception, harmonise the InfoSoc research exception across the EU and introduce a binding EU-wide secondary publication right; contractual circumvention of research exceptions must be ineffective. Knowledge is infrastructure – access to it decides participation, including in the 2,500 written languages for which GISAD’s companion project getmySense develops translation capability.

GISAD – Global Institute for Structure relevance, Anonymity and Decentralisation · Registered in the EU Transparency Register, No. 244298340978-40

Sources: European Commission – Have your say: Targeted initiative for a better copyright environment for European creativity and innovation · European Commission – Call for evidence on the review of EU copyright rules · European Commission – Targeted consultation on the effect of technology on copyright