Munich: Copyright Protection for AI-Generated Works

The case concerned a fundamental question: whether a logo created using generative AI software is entitled to copyright protection under Section 2 of the German Copyright Act (UrhG). The plaintiff sought an injunction against the defendant’s use of graphic symbols that the plaintiff had created using generative AI.

The central question before the court was whether the user’s involvement in the creative process — through the provision of textual instructions (prompting) — is sufficient to define them as a “creator” whose personality is reflected in the final product, thereby rendering the product a “work” (Werk) protected by law.

The Court’s Key Findings

In its decision, the court reaffirmed the foundational principle that only a “personal intellectual creation of a human being” is entitled to legal protection. It held that the mere use of software or technology does not automatically preclude protection, but that protection depends entirely on the degree of genuine creative influence exercised by the human over the process, notwithstanding the use of the software. To this end, the court defined the conditions for recognising copyright in AI-generated outputs:

The existence of human influence during or after the prompting stage; a requirement that the final product reflect the user’s creative choices and unique personality; and the existence of “creative dominance,” such that the elements introduced by the human are so significant that the product is identifiable as their original work rather than a technical output of the machine. The court quoted from its ruling (unofficial translation):

“The plaintiff is not entitled to the asserted claims under Section 97(1) of the German Copyright Act (UrhG). The disputed products are not works of applied art protected by copyright under Section 2(1) No. 4, (2) UrhG… the mere selection of an AI product from a number of ‘suggestions’ is not sufficient in itself.”

In analysing the specific facts, the court examined the prompts provided by the plaintiff and found that they constituted merely technical instructions or result-oriented requests — such as requests to “soften the facial features” or “make the hands more realistic.” The judge held that these instructions did not rise to the level of artistic decisions expressing the creator’s personality; rather, the plaintiff had left the work of aesthetic design to the AI and had acted only as an “editor” or “commissioner”, not as a creator.

“Figuratively speaking, it is required that ‘the use of the AI model be closer to an assistant than to an independent creative tool… Copyright does not reward and protect investment, expenditure of time or diligence, but only and exclusively the result of creative activity.”

Munich vs. Beijing: Contradictions and Insights

The claim was dismissed after the court found that the logo in question lacked “the character of a work” (Werkcharakter), because the technical activity of the machine clearly predominated over the plaintiff’s limited human creative contribution. The ruling sets a high bar for AI users, clarifying that prompt engineering alone will generally not attract protection, unless it involves a complex and unique process of personal selection and design. The decision aligns with rulings issued in the United States — for example in the Zarya of the Dawn case — and reinforces the policy against automatically protecting raw AI outputs, while providing tools for distinguishing between general technical instructions and protected artistic design.

While the Munich court adopted a conservative and restrictive line, the Beijing Internet Court in the Li v. Liu case took a surprisingly opposite approach. In that case, it was held that an image created using Stable Diffusion is entitled to copyright protection. A comparison of the two rulings yields several critical insights:

1. The prompt as a vector and evidence of intellectual effort – the Chinese court treated the prompt as a “significant intellectual contribution” and proof that the human determined the composition and aesthetics. In Munich, by contrast, the prompts were viewed primarily as technical instructions for “refining” an existing result. In this respect, scholars have recently proposed examining the level of creativity and originality in the prompt itself, which may be the key to distinguishing between a “technical instruction” and “personal expression” — with unexpected and unique expressions attracting a higher originality score that would justify broader protection.
2. Economic incentives versus protection of the creator’s personality – the German ruling is faithful to the continental tradition (Droit d’auteur), which enshrines the bond between a work and the soul and personality of its creator. The Chinese ruling, by contrast, reflects a more utilitarian approach: the Beijing court explicitly noted that recognising copyright in AI outputs would encourage people to use these tools and advance the technology industry. There is here a collision between the value of morality and the human on the one hand, and the value of progress and efficiency on the other.
3. Defining the ‘creator’: director versus clerk – in China, the user was regarded as a kind of “director” who guides the machine toward a desired result through a series of selective choices. In Germany, in almost identical circumstances, the user was treated as a kind of “client” making requests of a professional (the machine). This contradiction shows that the definition of the concept of “creative choice” is today subjective and depends on the judge’s interpretation of the question of where the “instruction” ends and the “work” begins.
4. The need for international uniformity – the contradiction between the rulings creates a state of global legal uncertainty: the same image may be protected in China but freely available to all in Germany or the United States. This situation reinforces the need for new international conventions to regulate the status of AI-generated works, since the internet does not respect these jurisdictional boundaries.

In conclusion, the conflict between the Munich ruling and the Beijing ruling illustrates that the legal world is at a point of bifurcation: whether to protect the “process” (the human effort in operating the AI) or to continue protecting only the “source” (the direct human execution). The winner of this contest will determine what the creative economy of the 21st century looks like.