AI and Copyright Law in Germany – Legal Certainty in Handling AI-Generated Content

The central problem with AI-generated content is that German copyright law (§ 2 Urhebergesetz; UrhG – the German Copyright Act) only protects works that are products of “original intellectual creation”. In German, the keyword is “persönliche geistige Schöpfung” (literally “personal intellectual creation”), which refers to works that bear the individual stamp of their human creator. This is the fundamental criterion that any work must fulfil to qualify for copyright protection under German law.

According to established German case law, this concept requires both creative individuality and a human author. The work must reflect the creator’s personality and creative choices. AI-generated content regularly fails to meet both criteria. It is created through statistical pattern recognition and algorithmic processes without the kind of creative individuality that copyright law recognises. It also lacks a human creator as the legal subject capable of holding rights.

This has concrete consequences for copyright classification:

• Lack of Work Quality: AI-generated texts or algorithmically created images typically do not reach the required threshold of creativity (“Schöpfungshöhe” – literally “height of creation”). This German legal concept refers to the minimum level of creative originality required for copyright protection. The creator’s personality must manifest in the work itself through individual creative choices. This personal expression must emerge during the creative phase of the creation process.
• Distinguishing AI as a “Tool” from AI as an Autonomous “Creator”: If AI is used merely as an assistive tool within the user’s own creative process, the user generally maintains creative control. They may potentially claim copyright protection. In contrast, if the user simply provides instructions for autonomous AI generation, they will likely be unable to assert rights under the German Copyright Act (UrhG). Their role resembles that of a commissioning party that merely establishes framework conditions. The actual creative process is executed by the AI itself.
• Issues Surrounding Co-Authorship: Even when users invest significant effort in crafting prompts, it is doubtful whether they can be considered co-authors under Section 8 UrhG (German Copyright Act). German law recognises co-authorship only when multiple people collaborate in creating a work where their individual contributions cannot be separately exploited. However, prompt creation typically involves providing conceptual input or instructions rather than directly shaping the specific creative expression that copyright law protects.
• Legal Status of AI-Generated Content: Whether AI-generated content qualifies as a protected work meeting the required threshold of originality must be assessed on a case-by-case basis. In most instances, however, this will not be the case due to the absence of human authorship and personal intellectual creation. As a result, such content is considered to be in the public domain. This means anyone can freely use, reproduce, and modify it without infringing any usage rights (the economic rights that allow copyright holders to control how their works are exploited commercially).

German copyright law (the “Urheberrechtsgesetz” or UrhG, enacted in 1965 and regularly updated) does not currently contain specific regulations for AI-generated works. The law was designed for human creativity and has not yet been adapted for the AI era.

The current legal framework essentially depends on the definition of the term “work” regulated in Section 2 UrhG, which requires human authorship and creativity. This is complemented by regulations about the transfer of rights and rights of use in Section 31 UrhG, which governs how copyright holders can license their works to others.

This creates a legal gap: while the law clearly states that AI cannot be an author, it provides little guidance on the complex practical questions that arise when humans and AI collaborate in creative processes.

The AI Act adopted in 2024 is currently setting the standard at the European level. This legislation creates binding rules for handling AI, though without directly modifying copyright law. In parallel, the EU Commission is discussing models to ensure compensation for the use of protected works in AI training processes.

Companies must prepare for a continuously evolving legal landscape that will require ongoing adaptations. Decisions regarding potential adjustments to copyright law in relation to AI are primarily made at the European level. The key framework here is the relevant EU Copyright Directive, which is scheduled for review and evaluation in 2026. At that point, if not sooner, it will be necessary to assess whether existing copyright provisions remain viable given technological developments.

Furthermore, the German Federal Ministry of Justice (BMJ – “Bundesministerium der Justiz”, the federal government department responsible for legal policy) is advocating for the inclusion of “AI and Copyright” in the new EU Commission’s work programme for 2024-2029. According to the BMJ’s position, there is a need to promptly and openly examine whether AI-generated content requires an adapted legal framework based on EU law.

Using copyrighted works for training AI systems raises substantial legal questions that are of central importance to companies. Whether reproducing copyrighted content for machine learning purposes is permissible currently falls within a legal grey area. However, legislators have implemented a protective mechanism through the so-called “opt-out” procedure, sometimes referred to as “machine-readable reservations”.

Rights holders can prohibit the use of their works for AI training purposes through an express reservation. For instance, they can include appropriate notices on their websites or in metadata. This might take the form of specific tags, robots.txt files, or clear legal notices. Once an opt-out declaration has been made, using the relevant content for AI training becomes legally impermissible. This regulation requires companies to carefully review their training datasets. Failing to observe existing usage reservations can result in significant liability consequences under copyright law, including claims for damages and injunctive relief (court orders to stop the infringing activity).

When using AI-generated outputs, the recognisability of pre-existing works becomes the decisive criterion for copyright assessment. If protected works are identifiable within AI-generated content, any act of use requires prior consent from the rights holder. This includes reproduction, distribution, or public communication. This also encompasses adaptations and transformations (derivative works), unless sufficient distance from the original work is maintained to constitute a new, independently protectable creation. The content can generally be used freely when works are no longer identifiable. However, drawing these distinctions requires nuanced case-by-case analysis, as the threshold for recognisability is not always clearly determinable.

The legal landscape surrounding artificial intelligence and copyright is evolving rapidly. For companies, this means those deploying AI systems must understand and actively mitigate legal risks. Mistakes can lead not only to financial damage but also to significant reputational harm. A well-thought-out strategy is therefore essential:

• Careful Contract Design: The foundation for legally compliant AI use lies in clear agreements with providers. Only through precisely defined licensing and usage rights can companies prevent third-party claims. Companies should particularly examine whether generated content can be used exclusively.
• Comprehensive Risk Assessment: Additionally, a thorough risk assessment of deployed AI systems is indispensable. Companies should obtain a complete overview of all AI systems in use, evaluate associated risks, and identify the corresponding legal obligations they must fulfil under applicable regulations.
• Labelling Requirements: Special attention must be paid to mandatory labelling obligations. Deepfakes and texts informing the public about matters of public interest must be marked as AI-generated. Similarly, users of AI chatbots must be informed that they are interacting with artificial intelligence.
• Minimising Liability Risk: Finally, liability risk can be substantially reduced through consistent compliance with legal regulations, such as the AI Act, and by fulfilling all due diligence obligations (the standard of care that a reasonable business would exercise in similar circumstances). A proactive compliance strategy protects companies from legal consequences while simultaneously enabling the safe utilisation of AI technologies’ potential.
• AI Guidelines and Training: Implementing AI training programmes and internal AI guidelines represents another crucial building block. Leadership, department heads, and all employees should be trained in the responsible use of AI tools. A binding AI policy provides company-wide regulation for the appropriate deployment of these technologies.