pdf

Keywords: Copyright, Artificial Intelligence, Training Data, Text and Data Mining Exception

The defendant, LAION e.V., developed a data set containing almost 6 billion image-text pairs, accessible to the public at no charge for AI training.
- This data set includes hyperlinks to images and textual descriptions for images available on the Internet.
- The data set aims to assist in training generative artificial intelligence.
- The images were sourced from a third-party dataset, which included URL links and textual descriptions.
The defendant downloaded images, checked textual descriptions against images, and filtered out mismatched entries.
The data set included at least one image copyrighted by the plaintiff, obtained via a photo agency's website that had specific usage restrictions.
As part of this process, an image copyrighted by the plaintiff and made available online via the website of a photo agency was recorded, downloaded, analysed and included in the new data set with its metadata.
The photo agency had issued a usage reservation in English in its terms of use, according to which visitors to the site were prohibited from "downloading" or "scraping" content from the site using automated programs.

The plaintiff demanded that the defendant refrain from reproducing the plaintiff's image for the creation of AI training data sets in the future.

The plaintiff argued that the defendant must cease using his image for AI training datasets, citing copyright protections.
The case centered around the legality of downloading the copyrighted image under the German Copyright Act.

Resolution: The Hamburg Regional Court dismissed the action and found that the defendant's actions fell under the copyright exception for text and data mining (Section 60d of the German Copyright Act).
Text and Data Mining Definition: Automated analysis of digital works to extract information about patterns, trends, and correlations.
The court concluded that the defendant's comparison of image content with existing descriptions was permissible under this exception.
Defendant’s source website used a natural language reservation, which the Chamber opined that it could be considered in a machine readable technology.
The plaintiff failed to carry its burden of proving that the exception did not apply.

Although the Chamber did not need to determine whether the general exception for text and data mining (Section 44b of the Copyright Act) was also available to the defendant, it offered obiter dicens on its potential application.
The general exception for text and data mining – unlike the more specific exception for text and data mining for the purposes of scientific research – permits the rights holder to reserve the use of its work for text and data mining through an express declaration.
A significant point of discourse was the effective communication of reserved rights on works online:
- For online works, a machine-readable reservation of rights must be utilized.
The reservation on the photo agency’s website was deemed ineffective as it was in natural language rather than machine-readable form.
The Chamber opined that the meaning of “machine-readable” should be assessed in light of the technology available at the time that the copyrighted work was reproduced.

The court did not resolve the legality of future uses of the image in AI training.

German Copyright Act Sections: 44a, 44b, 60d
EU Legislation:
- Arts. 3-4 of Directive (EU) 2019/790 (EU Digital Single Market Directive)
- Art. 5 of Directive (EC) 2001/29 (InfoSoc Directive)
- Art. 53(1)(c) of Regulation (EU) 2024/1689 (AI Act)

The case reflects the evolving interface between copyright law and technological advancements, particularly in the realm of AI and data mining practices.