Five years after the EU's Digital Single Market Copyright Directive entered into force, the European Writers' Council (EWC) says the law has failed to deliver fair pay and transparency for the 220,000 professional writers and translators it represents. In a position paper released June 1, 2026, the group called for urgent revisions to address what it describes as persistent implementation gaps and a crisis around generative AI training data.
What the Directive got right
The EWC acknowledged several tangible wins. The mandatory text and data mining exception for research organizations has accelerated scientific collaboration. The out-of-commerce works mechanism allowed cultural heritage institutions to digitize previously inaccessible collections. Most visibly, Article 17 forced major online content-sharing platforms to seek licenses or block infringing material. This led to more licensing deals and some new revenue for creators, along with greater public awareness of creative work's value.
Where it fell short for working authors
The transparency and remuneration provisions have not functioned as intended. Article 18, which calls for appropriate and proportionate pay, remains a vague principle rather than an enforceable right. Contracts stayed opaque and advances stayed low across much of the industry.
Article 19's transparency obligation has been undercut by confidentiality clauses - often imposed through contractual pressure - that prevent authors from auditing how their work is used. The contract adjustment mechanism in Article 20 rarely gets used because individual authors cannot shoulder the legal costs and evidentiary burden required to pursue a claim.
On the platform side, smaller services have applied Article 17 with excessive caution, leading to over-blocking of legitimate content like quotations, parodies, and user-generated material. Enforcement varies widely between member states, creating a fragmented system that hits cross-border authors hardest.
The AI training data problem
The most urgent issue, according to the EWC, is the interaction between the Directive and generative AI. Article 4 permits text and data mining for any purpose, subject to a rightsholder opt-out. In practice, the group said, AI developers have largely ignored or circumvented those opt-outs. Authors are watching their work train large language models without consent, attribution, or payment.
"The 'opt-out' system must be legally binding and technically operable," the EWC wrote. The organization wants the European Commission to issue clear guidelines requiring AI developers to report on training data usage, making any use of copyrighted works traceable and compensated. The group also called for a Generative AI and LLM regulatory framework that explicitly confirms AI training constitutes reproduction requiring authorization, with machine-readable opt-outs that are automatically respected and the burden of proof placed on AI providers.
Why this matters for creatives
The EWC's assessment confirms what many working writers and translators have experienced: a law designed to rebalance the digital market has not shifted power or income in their favor. Without enforceable remuneration floors, standardized reporting formats, and binding AI training rules, individual creators remain in a weak bargaining position. The group expects a formal revision of the Directive during the next Commission term and plans to propose specific legislative amendments. For professionals whose livelihoods depend on copyright, the next round of policymaking will determine whether the legal framework catches up to how their work is actually being used - and whether they get paid for it.
Your membership also unlocks:
