Lawful media monitoring in 2026 is not defined by any single law, but by the interplay of copyright and AI regulation. Anyone analysing content automatically has to satisfy both levels: the question may I process this content? and the question how transparent and safe must my AI system be?
Two frameworks, one field
In the debate, the EU AI Act and copyright are often treated separately. In the practice of media monitoring they mesh together. Copyright — specifically the TDM rules of the DSM Directive — governs whether content may be analysed automatically at all. The AI Act governs the transparency and due-diligence requirements the AI system in use must meet. Only together do the two form the complete legal framework.
We consider this the most important shift of recent years: compliance is no longer a matter of ticking a box, but of an end-to-end, traceable processing chain — from the source to the generated summary.
The copyright side: TDM and the reservation
Under Art. 4 of the DSM Directive and its national transpositions — §44b UrhG in Germany, §42h öUrhG in Austria — text and data mining is permitted in principle, as long as the rightholder has not declared a machine-readable reservation of use. This seemingly simple rule has a demanding consequence: those who capture content must actively evaluate and respect opt-out signals, not merely ignore whatever is technically reachable.
For us, that leads to a clear stance: capture only from lawfully accessible sources and consistent respect for declared reservations of use. "Technically possible" is not the same as "legally permitted".
The AI-law side: transparency as an obligation
The EU AI Act introduces a risk-based approach and imposes transparency requirements on generative AI in particular. For providers that automatically produce summaries from media content, this means above all: traceability. Generated content should be recognisable as such and traceable back to its underlying basis.
The European Commission has designed the AI Act as a framework that applies in stages, with its obligations entering into force gradually. The authoritative text is decisive; the European Commission's overview of the regulatory framework for AI offers a starting point.
Our consequence
At mediaintel, every synthesis is anchored to the substantiated statements of the original sources and traceable back to them — no freely formulated claims. This architecture grew out of conviction, but increasingly aligns with what regulation demands.
What this means in practice
For organisations that buy in or operate media monitoring, three test questions can be derived:
- Provenance: Is content captured lawfully and in line with reservations?
- Transparency: Are automated analyses and summaries traceable and recognisable as AI-generated?
- Data protection: Is the processing of personal mentions handled in a GDPR-compliant way?
Anyone who can answer these three questions cleanly is well positioned for 2026 — regardless of how the details of regulation evolve. The direction is clear: away from blind scraping, towards a substantiated, transparent processing chain.