Penguin Random House filed a copyright infringement lawsuit against OpenAI in a Munich court on March 31, alleging that ChatGPT reproduced content from the German children's book series Coconut the Little Dragon that was "virtually indistinguishable" from the originals [1]. The suit was filed against OpenAI's Irish-based European subsidiary.

The case turns on memorization. When prompted to write about the dragon protagonist, ChatGPT generated a narrative story, cover art depicting the orange dragon and companions, back-cover blurb text, and step-by-step self-publishing instructions -- all closely matching author Ingo Siegner's creative work [2]. Penguin Random House argues this demonstrates the model did not learn patterns from the material but memorized it wholesale during training. Der kleine Drache Kokosnuss spans more than 30 volumes, with television adaptations and two films, making it one of the most valuable children's properties in Germany [1].

The suit is the first major AI copyright case filed in a European court and tests a distinct legal theory from the US lawsuits brought by authors and publishers against OpenAI [2]. The American cases argue training itself is infringement. The Munich case argues the output is infringement -- that what ChatGPT produces is a copy, not a creation. If Penguin Random House prevails, the precedent would establish that AI-generated content can constitute direct copyright violation under European law.

-- ANNA WEBER, Berlin