The EU Copyright Review: A Make-or-Break Moment for European AI

What is happening?

The European Commission has launched a combined Call for Evidence regarding two connected, but separate issues: the review of Directive 2019/790/EC on Copyright in the Digital Single Market (the ‘CDSM Directive’) and a potential upcoming targeted legislative initiative aimed at fostering a better copyright environment for European creativity and innovation. The process, which also includes an externally commissioned study [LINK], seeks stakeholder feedback to address market and technological developments, in large part linked to artificial intelligence (AI), but also other issues such as performer remuneration and online piracy.

Rooted in Article 30(1) of the CDSM Directive, which mandates a formal review of the legal framework no sooner than June 2026, the consultation focuses on several major issue areas:

• The challenges generative AI poses to rightholders regarding content control, licensing models, and remuneration – including the rise of AI-generated personal characteristic imitations, known as “impersonifications” of performers. This part is of the highest importance for the EU innovation ecosystem, as it has the potential to severely disrupt any AI training in the region. 
• Persistent online piracy affecting live content and sports events.
• Cross-border imbalances in single equitable remuneration for third-country musical works under the legacy of the CJEU RAAP judgment (Case C-265/19).
• Harmonization fragmentation concerning the optional scientific research exception under Directive 2001/29/EC (the ‘Infosoc Directive’) and obstacles to secondary publication rights for publicly funded research. These issues also having direct and profound implications for European innovativeness and development of AI. 

What are the implications?

From the perspective of the European AI ecosystem, this consultation represents a potential legal risk and may contribute to legislative changes that could severely impede or completely halt regional deep-tech innovation. The primary risks include:

• Undermining the current legal underpinnings of AI training in the EU: The European AI development sector relies heavily on the Text and Data Mining (TDM) exceptions established under Articles 3 and 4 of the CDSM Directive. Any regulatory attempt to narrow, carve out, or restrict these exceptions creates an immediate existential threat to model training capacity within the Union. The consultation document implies this is the aim and references directly the JURI report on Copyright and Generative AI, which contains ideas very detrimental to the future of AI training (see our commentary HERE  and HERE)
• Prohibitive Transaction and Legal Costs: Shifting away from the current opt-out regime towards mandatory licensing regimes, heavily restricted accessibility, or complex ex-ante transparency standards will create an unmanageable administrative burden and drive up costs. While deep-pocketed foreign technology incumbents can absorb these compliance costs, European startups, scale-ups, and open-source initiatives will be crushed under “legal spaghetti” and upfront costs of AI training, before any revenues appear on the horizon. 
• Paralysis through legal ambiguity: Expanding rights control mechanisms or introducing vague criteria for “appropriate remuneration” before clarifying technical feasibility introduces immense litigation risks. If developers face continuous threat of ex-post enforcement or algorithmic “machine unlearning” mandates that are economically unviable, capital and top scientific talent will flee the EU. This risks turning Europe into a permanent net importer of foreign AI foundational models. This is already the case in the EU, where traditionally risk-averse companies are being subjected to pressured not to use publicly available data and strong anti-AI rhetoric is abundant in the mainstream media. 

While these outcomes are of course not guaranteed, many clues point in the direction – both in the consultation document text itself (which, while noticing that there needs to be consideration given to the access to copyright-protected content) make it clear that the main aim of the upcoming proposals will be to find “ways to enhance the licensing and enforcement of copyright and related rights in the AI context” and “improve the conditions for creators’ remuneration”.

What is our opinion?

The AI Chamber acknowledges the Commission’s intent to review the CDSM framework and we agree with the need to adapt rules to changing circumstances. We also agree with the need to improve the scientific exception, in particular by making it easy and risk-free for researchers to commercialise their findings. We must however raise serious public policy concerns regarding the underlying philosophy of this consultation when it comes to rules governing AI model creation.

Critical Assessment of the Preemptive and Defensive Stance

The consultation document demonstrates a highly problematic, structurally defensive bias that treats generative AI primarily as an exogenous threat to control and subject to more stringent remuneration requirements. Alarmingly, the Commission goes as far as to preemptively declare that the newly enacted AI Act and subsequent copyright implementation measures “may not be sufficient” to protect rightholders, doing so before full empirical evidence from the consultation is collected or evaluated, and while the horizontal AI Act rules are still actively being phased in across Member States.

By fixating purely on theoretical ex-ante risks and procedural friction, the text completely overlooks the immense horizontal efficiency gains AI generates for the broader economy, material science, creative productivity and society writ large. AI promises to improve not just productivity, but people’s lives and health. We shouldn’t forego these opportunities in order to make it easier for narrow financial interests to be enforced. 

The text also extensively mentions “generative AI” as the target, overlooking the many other types of AI that also relies on access to data, but has no direct connection to the “cultural and creative sectors” this initiative aims to protect. 

Contrast with Global Competitors

This overbroad precautionary approach, apparent in the document, stands in stark contrast to the strategic agendas deployed by global technology superpowers: the United States, China, Japan or Israel. If we do not ensure a level playing field for our AI creators and researchers we will once again lose our best and brightest, who will simply relocate to more favourable jurisdictions. 

Call to action and recommendations

To safeguard the competitiveness of the European innovation sector and prevent total deindustrialization in the digital sphere, the AI Chamber demands that the co-legislators implement the following parameters in the upcoming CDSM review and targeted legislative instrument:

1. Firmly protect, expand and improve Articles 3 & 4 CDSM: The text and data mining exceptions must be robustly safeguarded against any legislative narrowing down. The EU must provide an absolute, harmonized legal presumption that transient data use for machine learning training does not constitute a copyright infringement.
2. Reject Double Regulation: Avoid layering sector-specific copyright transparency or data audits that create conflicting overlaps with the commitments already mandated by the AI Act and the horizontal Data Act framework. Compliance tools must remain simple, standardized, and fully compatible with existing technical standards.
3. Prohibit National Carve-Outs and Gold-Plating: The CDSM Directive review must move toward a logic of maximum, exhaustive harmonization. Individual Member State deviations or national judicial narrowing of copyright  must be explicitly prohibited to prevent single market fragmentation. This concerns both the TDM and scientific research exceptions. 

Call to Action:

This consultation is a big step in a process that may become the defining moment for the European digital sector. We call upon all tech companies, AI professionals, general counsels, founders, and researchers to actively participate in this consultation. We cannot allow legacy business models to unilaterally write the rules of Europe’s technological future. The legacy copyright industries are already heavily invested in the lobbying campaign that has already borne fruit in the form of the JURI report. We cannot stand by idly or intervene at the last moment, when the decision will already be made. 

We all can and should have a say. Let’s shape a pro-innovation Europe together. Access the consultation HERE and HERE