EU lawmakers push forward an AI Act draft, which includes requiring companies deploying generative AI tools to disclose copyrighted material used for training
Learn about the 5 errors to avoid … Sabrina Ortiz / ZDNet : Generative AI might soon face some major copyright limitations from the EU Sam Schechner / Wall Street Journal : Europe to ChatGPT: Disclose Your Sources Sharon Goldman / VentureBeat : EU lawmakers pass draft of AI Act, includes copyright rules for generative AI Luke Jones / WinBuzzer : EU to Launch New AI Act with Nuanced ChatGPT Classifications Joe Fay / DEVCLASS : Brussels greenlights AI Act, chews over Cyber Resilience legislation Finextra : Generative AI governance targeted in proposed EU rules Paul Rainford / Light Reading : Eurobites: Bouygues, SFR want to extend network sharing to 5G sites Ryan Morrison / Tech Monitor : EU says generative AI makers must declare copyrighted content Agne Cimermanaite / Metaverse Post : EU Reaches Early Agreement on AI Act with Focus on Transparency and Risk Mitigation Sarasvati NT / MediaNama : European Union to introduce new Copyright rules for generative AI tools in its AI Act LinkedIn: Dazza Greenwood : This is interesting. Apparently, under the draft EU's AI Act, “a subcategory of ‘General Purpose AI’ that includes tools such as ChatGPT” … Scott Sholder : EU's moving on AI legislation. Interesting that generative AI platforms will need to disclose if they used copyrighted materials in their systems. … Rupert Breheny : The regulatory environment is finally catching up. Clear visibility on copyrighted materials in model training will be key for companies … Mark Milstein : After months of negotiations and two years after draft rules were proposed, EU lawmakers have reached an agreement and passed a draft of the Artificial Intelligence … Stefan Harrer : This is big - milestone alert in AI governance and regulation: today, after months of negotiations and two years after draft rules were proposed … Carissa Véliz : “Companies deploying generative AI tools, such as #ChatGPT, will have to disclose any copyrighted material used to develop their systems … Tweets: Mario Guglielmetti / @mario_gug : “Companies deploying generative AI tools, such as ChatGPT, will have to disclose any copyrighted material used to develop their systems” 👏 Still what about personal data? Special categories? GDPR applies, however do we know? wrt Google Bard for instance? https://www.reuters.com/... Tony Roberts / @phat_controller : Under the European AI Act tools will be classified according to their risk level. Areas of concern expected to include biometric surveillance, spreading misinformation or discriminatory language. https://www.reuters.com/... Siddarth Pai / @siddarthpaim : Proposed EU legislation would require AI developers to list copyrighted material used in tools such as ChatGPT https://www.wsj.com/... Soon Web will be transitioning to high grade end point security which would not allow any kind of web scrapping , automation tools etc Matthew Lamons / @mlamons1 : The rules for copyright surrounding Artificial Intelligence have thus far been loose, confusing, or nonexistent. The EU has proposed a new set of copyright rules just for generative AI. #AI #ML #futurism #IntelligenceFactory #digitaltransformation #DX https://www.reuters.com/... Jarrett Catlin / @jarrettcatlin : The first major bill to pass will be the benchmark all others are compared to Just like how CA modeled their data laws after GDPR https://twitter.com/... Dr. Ikramul Haq / @drikramulhaq : Companies deploying generative AI tools, such as ChatGPT, will have to disclose any copyrighted material used to develop their systems, according to an early EU agreement that could pave the way for world's first comprehensive laws governing technology. https://www.reuters.com/... See also Mediagazer
Context & Ripple Effects
The draft put copyright provenance alongside the EU’s broader risk-based AI framework, making generative models a distinct regulatory problem rather than just another software category. It also arrived amid GDPR-focused scrutiny of model data practices, including questions over scraping and deletion requests.
The proposal became an early fault line in the Act’s negotiations: foundation-model rules were still unresolved later that year before lawmakers reached a final AI Act deal. That arc matters because it shows training-data disclosure moving from a draft demand toward an implementation issue.
First-order effects
- Companies deploying generative AI in the EU face pressure to identify and document copyrighted training material, turning data provenance into a compliance workstream rather than a purely technical or legal-records issue.
- Copyright owners gain a clearer basis to ask how generative systems were trained, while providers must prepare for greater scrutiny of the disclosures they can make.
Second-order effects
- Model developers and deployers are pushed to improve contracts, dataset records, and supplier assurances, especially where training data passes through multiple intermediaries.
- The disclosure approach raises the value of licensed, traceable content and increases pressure on firms whose models rely on poorly documented web-scale data; it parallels later U.S. legislative interest in training-data disclosure.
Third-order effects
- If disclosure becomes a durable enforcement norm, AI competition may increasingly hinge on rights-cleared data access and auditable data supply chains, not only model performance.
- The AI Act’s handling of foundation models suggests that general-purpose systems will be governed through ongoing documentation and accountability requirements; the later voluntary compliance code points in that direction, though the practical burden depends on final rules and enforcement.
The trend: Generative AI governance is shifting from broad principles toward traceable obligations around the data, documentation, and rights behind model outputs.