THE BOTTOM LINE
- You Control Your Content: You have the legal right to prohibit AI developers from using your publicly available website content (text, images, data) to train their models. Simply being online does not grant a free pass for data mining.
- Action is Required: This protection is not automatic. To legally prevent AI training on your data, you must explicitly “reserve your rights” in a clear manner, for example, within your website’s terms and conditions or in a machine-readable format.
- Risk for AI Developers: AI companies must now actively check for and respect these opt-outs. Ignoring a clearly stated reservation and scraping the data for training purposes constitutes copyright infringement, creating a significant litigation risk.
THE DETAILS
This landmark judgment from the EU’s highest court provides much-needed clarity at the intersection of copyright law and artificial intelligence. The case revolved around the “Text and Data Mining” (TDM) exception within the EU Copyright Directive, which was designed to foster research and innovation. The core question was whether this exception gave AI developers free rein to use any lawfully accessible online data for training models, or if copyright holders retained the power to say “no.” The Court has come down firmly on the side of the rights holder, establishing a critical “opt-out” precedent for businesses across the European Union.
The Court’s reasoning emphasizes that the TDM exception was intended to be a balanced tool, not a blanket override of copyright. It clarified that while the exception allows for mining of lawfully accessed content, it is subject to a crucial condition: that the rights holder has not “expressly reserved” the right to do so. The judges reasoned that merely making works available on the internet—for instance, on a corporate website or a publisher’s portal—is not an implicit license for third parties to use that content for commercial TDM activities like training a generative AI. The power to control one’s intellectual property remains with the creator.
For business leaders and their legal counsel, the message is clear: proactive IP management is essential in the age of AI. The ruling validates the practice of including explicit TDM restrictions in website terms of use and using machine-readable protocols (like robots.txt or metadata tags) to signal that content is off-limits for AI training. For companies developing or deploying AI, this decision elevates the importance of compliance. Their data acquisition processes must now incorporate robust checks to ensure they are not infringing copyright by training their models on content where the rights have been expressly reserved.
SOURCE
Source: Court of Justice of the European Union
