TDM Reform and Creativity: What Would It Mean for Europe’s AI Ecosystem

On Tuesday, 16 December 2025, SME Connect organized a webinar titled “TDM Reform and Creativity: What Would It Mean for Europe’s AI Ecosystem” bringing together policymakers, experts, and industry stakeholders to analyse ongoing policy debates around Text and Data Mining (TDM) exceptions in EU copyright law and their practical effects on Europe’s AI innovation landscape. The webinar centred on how legal reforms to the TDM regime—particularly around Article 4 of the 2019 EU Copyright Directive and its opt-out mechanism—could affect AI training access, startup growth, industrial deployment, investment dynamics, and Europe’s global competitiveness in AI development, notably in comparison with the United States and China.

HORST HEITZ, Chair of the Steering Committee of SME Connect, opened the webinar emphasizing the timeliness of the debate and the need to give SMEs a practical understanding of possible policy directions. He highlighted communication gaps, noting that many SMEs sense upcoming changes without having clear information. He stressed that such uncertainty can already influence investment and location decisions. He repeatedly framed the discussion in terms of global competitiveness, warning that AI development is a race with no second chances and that Europe must not only reach the top but also stay there if it wants to secure its future economic position. Throughout the discussion, he underlined the need to balance innovation and competitiveness with fairness and the legitimate interests of the creative industries. He invited reflection on whether clearer rules, improved opt-out mechanisms, or broader remuneration approaches could offer solutions. He concluded by stressing that SMEs must remain visible in the policy debate and that Europe should pursue a pragmatic and inclusive path.

IVAN ŠTEFANEC, former MEP and President of SME Europe, provided a political perspective focused on concrete SME realities. He argued that TDM directly affects whether SMEs can lawfully access data, build AI-enabled products, attract investment, and compete globally. He illustrated this with examples from manufacturing, health tech, legal tech, and creative sectors. He explained that Article 4 was designed to allow TDM by default unless rightsholders opt out, but stressed that opt-outs are implemented inconsistently in practice. He warned that narrowing the exception or adding conditions would raise costs and legal uncertainty for SMEs. He concluded that more restrictive TDM rules risk pushing innovation out of Europe or concentrating it in fewer players, with negative effects across the value chain.

DANIELA WERBENIEC, Director of the Digital Forum at ZPP, focused on how SMEs perceive the debate and why uncertainty already has tangible effects. She explained that most SMEs do not follow copyright discussions in detail, but instead perceive growing risks around doing AI-related business in Europe. She noted that signals suggesting AI training may become legally risky, expensive, or fragmented create a chilling effect. In her view, the most damaging outcome is uncertainty over whether what is lawful today will remain lawful tomorrow. She confirmed that this uncertainty can influence location decisions even before any legal changes occur. She cited ElevenLabs as an example often referenced to illustrate how European-founded companies may scale elsewhere. She concluded that Europe must design TDM rules that allow SMEs to build AI responsibly and competitively.

MARCIN OLENDER, Director of Public Policy at the AI Chamber, argued that copyright policy may be the most decisive legal factor shaping Europe’s AI ecosystem. He warned that fear, uncertainty, and doubt can pressure companies into precautionary licensing and discourage innovation. He maintained that Europe’s rigid copyright framework has historically prevented certain products and services from emerging, while any reopening of Article 4 risks further restriction. He stressed that limiting access to training data also has cultural consequences, as AI systems would fail to reflect European content and values. He illustrated the real-world impact with examples of European talent relocating abroad and a sign-language AI project that faced major data-access barriers in Poland but not in the United States. He also addressed proposals for compensation or blanket licensing schemes, acknowledging that they may appear attractive in theory but warning that, in practice, they are difficult to administer, risk being controlled by narrow interests, and could ultimately hinder innovation rather than support it. He further cautioned that embedding copyright enforcement directly into AI systems could restrict creative freedom at the point of creation.

CAROLINE DE COCK, Head of Research at Information Labs and Senior Fellow at the Lisbon Council, highlighted that 2026 will be a critical year, as the Commission may reassess the Copyright Directive. She argued that moving from opt-out to opt-in or mandatory licensing would harm SMEs and fail to deliver the intended economic benefits. She stressed that regulatory stability is the most reassuring signal Europe could give companies, as uncertainty delays adoption and investment. Using the example of a small startup optimizing renewable energy grids, she explained why opt-in requirements would make many AI projects unviable. She warned that stricter TDM rules would entrench large foreign firms rather than constrain them. She pointed to Japan, which allows AI training as an analytical use distinct from enjoyment of copyrighted works, and Singapore, which has introduced a broad computational data analysis exception that cannot be overridden by contracts, as examples of alternative approaches that provide greater legal clarity while supporting innovation. She also addressed proposals for remuneration or compensation schemes, highlighting unresolved technical questions around how data use could be tracked, how funds could be collected, and how any resulting revenues could be distributed without creating excessive cost and friction.

BRIAN WILLIAMSON, author of the Communications Chambers report “AI, Copyright and the Public Good”, framed the debate around Europe’s long-term growth and creative capacity. He argued that AI will be a major driver of economic growth and a powerful creative and scientific tool. He recalled that copyright was originally designed to expand access to knowledge rather than restrict it. He emphasized that restrictions should focus on AI outputs, not on the inputs used for training. He described Europe’s current TDM framework as relatively permissive by international standards, though imperfect. He concluded with a pragmatic warning that reopening the TDM exemption risks producing a more complex and restrictive outcome, making stability the preferable option.

Speakers agreed that TDM reform is not merely a copyright issue, but one with far-reaching implications for Europe’s AI ecosystem, from research and startups to industry and investment. They emphasized the need to balance rightsholders’ interests with legal certainty, competitiveness, and innovation. Speakers underlined that Europe’s policy choices in this area will determine not only whether it can participate in the global AI race, but whether it can remain competitive over the long term. The webinar concluded with a call for continued dialogue, clearer communication, and stable regulatory frameworks to support both creative industries and Europe’s technological future.