At a Frankfurter Buchmesse panel on 7 October 2026, publishing and research specialists debated whether the EU AI Act’s transparency and copyright-related provisions give rights holders practical leverage. The amended regulation provides for administrative fines and gives the European Commission’s AI Office enforcement powers in specified cases; the panelists’ concerns focused on how effective the rules may be in practice.
Frankfurt panelists question practical deterrence
Elizabeth Crossick, RELX’s head of EU government affairs and global AI policy lead, characterized many requirements as weak and argued that limited consequences can undermine incentives to comply. Gerhard Lauer, a professor of book studies at Mainz University, said that recording every AI-assisted research step can be technically out of reach and burdensome for researchers. The panel also considered calls for stricter regulation alongside a more flexible approach.
That debate sits within a legal framework that has been in force since 1 August 2024 and applies in stages. Its scope includes specified providers and deployers, including certain operators based outside the European Union when an AI system’s output is used within it.
What the amended law provides
Regulation (EU) 2026/1744 gives the European Commission’s AI Office supervisory and enforcement powers within its remit. These include making non-compliance decisions, ordering corrective measures, imposing administrative fines and applying periodic penalty payments. The amended Act also provides for administrative fines.
The regulation sets two later application dates for high-risk requirements in Chapter III, Sections 1–3. For systems in Article 6(2) and Annex III, those requirements apply from 2 December 2027. For systems in Article 6(1) and Annex I, they apply from 2 August 2028.
Article 50 covers transparency in defined situations, including direct interactions between people and AI systems and certain AI-generated or manipulated content. For specified text published on matters of public interest, an exception applies when the text has undergone human review or editorial control and a person or organization holds editorial responsibility.
TRACE addresses scholarly-content provenance
TRACE—Trusted Retrieval & Attribution for Content Ecosystems—is a separate community initiative intended to support discovery and access, provenance and attribution, and usage reporting in scholarly AI communication. Provenance means information about where content came from and how it can be attributed.
A National Information Standards Organization (NISO) pilot within TRACE aims to develop a minimum metadata model for provenance at inference time—the stage when a model retrieves or uses information to produce an output. The pilot’s scope does not include copyright enforcement or attribution inside foundation-model weights.
What the Frankfurt Appeal requests
The Frankfurt Appeal calls for disclosure of the specific works used to train generative AI, fair licensing and effective means of enforcing copyright. These are advocacy requests from the appeal’s signatories.
