Penguin Random House has sued OpenAI, alleging ChatGPT violated copyright laws by reproducing “virtually indistinguishable” content from the beloved German children’s book series, “Coconut the Little Dragon” (Der kleine Drache Kokosnuss).
The lawsuit, which could reshape the legal landscape for generative artificial intelligence (AI) and content creation, filed last week in a Munich court against OpenAI’s Irish-based European subsidiary, centers on the chatbot’s ability to mimic the specific creative output of author and illustrator Ingo Siegner.
According to the filing, legal teams prompted ChatGPT to write a story about the titular dragon visiting Mars. The AI responded not only with a narrative but also with cover art, a back-cover blurb, and step-by-step instructions on how to upload the resulting manuscript to self-publishing platforms.
Penguin Random House argues that the AI’s ability to recreate the orange dragon and his companions is “clear evidence” that OpenAI’s large language models (LLMs) unlawfully “memorized” Siegner’s work during training. (“Memorization” refers to a phenomenon where AI systems store and reproduce long excerpts or specific visual details of their training data. While AI firms often argue this is a transformative process rather than traditional copying, publishers contend it is a direct theft of intellectual property.)
“Human creativity remains at the heart of our work,” said Carina Mathern, publisher for children’s and young-adult books at Penguin Random House. “While we are open to the opportunities offered by AI, the protection of intellectual property is our top priority. We are obliged to represent the interests of our authors.”
An OpenAI spokesperson said the company is reviewing the allegations, maintaining that they “respect creators and content owners” and are seeking “productive conversations” with publishers globally.
The lawsuit lands despite a January 2025 collaboration deal between OpenAI and Bertelsmann, the German media conglomerate that owns Penguin Random House. That agreement did not grant OpenAI access to Bertelsmann’s vast media archives, a boundary the publisher claims has been crossed by the AI’s training methods.
This is not OpenAI’s first legal hurdle in Germany. Last November, the Munich regional court ruled against the company in a case brought by the music rights society Gema, finding that ChatGPT had harvested protected lyrics to “learn” from popular artists.
The German Publishers and Booksellers Association hailed the lawsuit as a landmark step toward the regulation of generative AI. As “Coconut the Little Dragon” spans over 30 volumes, a TV series, and two films, the stakes for protecting such a lucrative franchise are high.
As Penguin Random House joins a growing list of authors and creators seeking judicial intervention, the case may provide a definitive answer on whether AI learning constitutes a fair use of the world’s literature or a high-tech form of piracy.
