Hamburg Court Delivers Ruling on the Use of Photographs to Train AI: No Copyright Infringement Has Been Found

Regional Court of Hamburg has made a significant decision in a case concerning the unauthorized copying of a photograph to create a dataset for generative artificial intelligence (AI) systems. Being one of the first court decisions in Europe regarding a dispute related to artificial intelligence makes this Kneschke v. LAION case crucial, as it touches upon issues such as whether AI data scraping constitutes text and data mining and, if so, how right holders might practically prevent AI developers from engaging in such activity.

The non-for-profit organization LAION was taken to court by photographer Robert Kneschke, claiming that LAION infringed on his copyright by copying his image without consent. The organization used the image for creating a dataset, which matches the images on the internet with text descriptive of those images. LAION as the defendant disputed the claims of copyright infringement and put forward that their activity falls within at least one of three exceptions to copyright, which are provided in German and EU law. The court ruled in favor of LAION and found that LAION’s activities do not constitute copyright infringement.

This decision is grounded in an exception outlined in the articles of Digital Single Market (DSM) Copyright Directive.[1] The ruling addresses the scope of temporary copying and text-data mining (TDM) exceptions and at the same time encourages further discussions on how to enforce copyright laws regarding AI.

Exceptions under the DSM Copyright Directive and their Application to the Case

The first exception cited in the case stems from article 5 of the Directive 2001/29/EC[2] which permits temporary reproductions that are transient or incidental, as a part to a technological process for lawful use without independent economic significance. In this case, the Court explicitly stated that the reproduction by LAION was “neither fleeting nor incidental”. Moreover, it stated that the images were deliberately downloaded in order to analyze them using a specific software. Thus, the defendant organization’s chance of relying on this exception was eliminated by the Court.

The second exception, Article 4 of the DSM Copyright Directive, allows for text and data mining (TDM) of lawfully accessible works with some restrictions. Reproductions can only be retained as long as necessary for TDM, and right holders must not expressly reserve their works from TDM use. After evaluating the facts of the case from different perspectives, the Court did not make a definitive ruling on whether creating datasets for AI training falls under Article 4, although significantly the Court explicitly stated that the TDM exception does apply to AI data scraping from wording within the EU AI Act. The AI Act requires providers of general-purpose AI (GPAI) models to comply with EU copyright law, specifically ensuring that they honor copyright holders’ rights to restrict the use of their works for text and data mining (TDM), as stated in Article 4 of the DSM Copyright Directive. In this regard, the Court underlined the requirement of the right-holders to state their intention clearly in a machine-readable format or in natural language to prohibit their works being used for TDM purposes. While leaving the questions unanswered about the Article 4 TDM exception for this particular case, the Court did find the LAION could rely on separate TDM exception that has been provided in German law, which also derives from the DSM Copyright Directive.

As a third exception, which is the reason for the dismissal of the case by the Court, Article 3 of the directive mandates that EU member states must establish an exception in their national copyright laws while allowing research organizations and cultural heritage institutions to reproduce and extract copyrighted works through text and data mining (TDM), where they have lawful access to those works for scientific research purposes. The dataset in question had been released to the public free of charge, and no evidence was presented that any commercial body had control over its operations. In this case, the Hamburg Court considered that LAION was a research organization, and that the creation of its data set was for the purposes of scientific research. The court stated that the dataset was made freely available to researchers, particularly in AI networks, which met the exception’s criteria. Therefore, the Court ruled that LAION’s actions were covered by section 60d of the German Copyright Act (Art 3 DSM), and as a result, there was no copyright infringement.

Possible Impact of the Judgment

This case holds considerable significance from a historical aspect. It represents the first legal examination and testing of the text data mining (TDM) exceptions outlined in the DSM Copyright Directive.

Although a local decision, the outcome of this case has the potential to not only effect how TDM is going to be interpreted and applied under EU law, but also has the potential to influence the future legal frameworks surrounding copyright and data use in the digital age. Testing of these exceptions in the court will potentially shape the relationship between copyright holders and AI developers, as impacting how data will be accessed, stored and used in various sectors. As AI systems increasingly rely on vast amounts of data for training and operation, the considerations and arguments set out in this case will play a crucial role in determining in a clearer way to which developers can access and use copyrighted materials for TDM purposes. These types of cases will ultimately reshape the balance between technologic innovations and copyright protection.

[1] For full text of the Directive (EU) 2019/790 of the European Parliament and of the Council of 17 April 2019 on copyright and related rights in the Digital Single Market and amending Directives 96/9/EC and 2001/29/EC see: https://eur-lex.europa.eu/eli/dir/2019/790/oj

[2] For full text of the Directive 2001/29/EC of the European Parliament and of the Council of 22 May 2001 on the harmonisation of certain aspects of copyright and related rights in the information society see https://eur-lex.europa.eu/eli/dir/2001/29/oj

Author: Stj. Av. Deniz Yegin