Introduction

Introduction

The use of artificial intelligence (AI) in the field of data mining confronts copyright law with new and complex issues. A central, but possibly not yet widely known standard in this context is §44b of the German Copyright Act (UrhG). This relatively new paragraph regulates the use of copyrighted works for text and data mining. In view of the rapid developments in AI technology, especially with advanced systems such as ChatGPT, Stable Diffusion or Midjourney, which can generate content independently, Section 44b UrhG is gaining practical relevance. It sets out the legal framework for the use of copyrighted works in the field of machine learning and AI development. This article examines the significance of Section 44b UrhG in the context of current technological developments and the associated legal challenges. In particular, the impact of these regulations on AI development practice and the need to continuously adapt copyright law to the constantly changing technological landscape will be discussed.

§44b UrhG: Permissibility of the training of AI systems

According to Section 44b of the German Copyright Act (UrhG), reproductions of legally accessible works on the Internet are permitted for the purposes of text and data mining, which is an essential basis for the training of AI systems such as ChatGPT. This arrangement allows such systems to be trained with a variety of existing texts or images in order to improve their pattern recognition and information processing capabilities. Despite its importance, however, this regulation is not generally known in practice, which leads to uncertainties and discussions, particularly with regard to the use of copyrighted works. These ambiguities can lead to legal gray areas, especially for the developers of AI systems and the authors of the content used. The question arises to what extent the use of such content for the training of AI systems affects the rights of the authors and whether additional measures, such as obtaining licenses or adapting the training methods, are necessary to avoid copyright infringements. In addition, the application of §44b UrhG raises questions regarding the distinction between permitted use and potential copyright infringements, which makes precise knowledge and interpretation of this regulation all the more important.

Right of objection and reservation of use

The §44b Abs.3 UrhG provides for a right of objection, which states that AI systems may only be trained with third-party works if the rights holder has not explicitly objected to such use. This right of objection enables website operators to prevent the use of their content by AI crawlers by storing a usage reservation in the robots.txt of their website. This option is an important measure to maintain control over your own content and prevent unauthorized use.

However, the problem becomes more complex when it comes to the manual, non-automated use of content to feed a proprietary AI. In such cases, it could be argued that the direct, conscious selection and use of content for the training of an AI represents a different quality of use than the automated reading by AI crawlers. This raises the question of the extent to which the right of objection under §44b para. 3 UrhG applies when a person or company specifically and manually selects content to train their own AI. This could be regarded as a more intensive form of use, which may not be covered by the provisions of Section 44b UrhG and therefore requires a separate copyright assessment.

In such cases, it may be necessary to obtain explicit permission from the author, especially if the content is used in a way that goes beyond mere reading and analysis for text and data mining. This raises further questions, for example in relation to the limits of copyright and the rights of authors, particularly when it comes to the use of content for the development and improvement of AI systems that may be used commercially. Clarifying these issues is crucial to finding a balance between the rights of creators and the needs of AI technology developers.

Conclusion

Section 44b UrhG is of central importance in the area of data mining using artificial intelligence. For creators and users of AI systems, especially for developers and providers of Software-as-a-Service (SaaS) solutions, a thorough understanding of this regulation is essential. It forms the legal basis for the use of copyrighted works in the context of machine learning and AI development.

As a lawyer specializing in copyright and IT law, my focus is on helping AI developers and SaaS providers to design their systems and processes in accordance with copyright requirements. This includes the implementation of mechanisms that ensure the copyright-compliant use of content, as well as informing and advising customers about the legal framework.

For SaaS providers offering AI-based services, it is particularly important to inform their customers about the legal limits and possibilities. This can be achieved through clear terms of use and, if necessary, by integrating compliance tools into the software.

Compliance with §44b UrhG not only offers legal security, but can also serve as a competitive advantage. In my role as a lawyer, I support companies in offering innovative AI solutions that both ensure copyright compliance and fully exploit their innovation potential.

In conclusion, I would like to emphasize that sound legal advice is crucial in this area. I offer comprehensive support to ensure that your products and services comply with current legal requirements. Together, we can turn legal challenges into opportunities and strengthen your position in the rapidly evolving world of AI technology.