From the Soul Life of the Creator to the Black Box of the AI Chatbot: How to Close the Attribution Gap of AI Output?

This article addresses the profound legal dilemma posed by Generative AI within the contemporary art scene, arguing that German copyright law must focus not only on the work’s inherent object qualities but in particular on the human’s demonstrable concrete control over the output. When artists input hyper-complex prompt chains into AI tools like Midjourney or DALL-E to generate a picture, the essential question arises: Who owns the result? This legal challenge is twofold: an attribution crisis and a transparency crisis. Because the AI autonomously produces works without the required link to a human author’s personality, it creates a fundamental attribution gap. The challenge is no longer merely about the work itself, but about establishing a verifiable human influence amid the AI’s ‘black box’ processes. This analysis explores this critical question by examining the chain of attribution across three distinct stages of human influence over the AI process: Architecture/Training Data (Upstream), Prompting (Direct), and Post-processing (Downstream).