MEDIANAMA
Technology and policy in India
Penguin Random House has sued OpenAI, alleging its chatbot reproduced content from a popular German children’s book series.
What triggered the lawsuit: The case involves the children’s book series Coconut the Little Dragon by Ingo Siegner. Penguin Random House asked ChatGPT to generate a story about the dragon set on Mars. The output allegedly included text similar to the original, illustrations of the same characters, a book cover, a back cover blurb and instructions for self-publishing. The publisher claims the result was “virtually indistinguishable” from the original work.
“Memorisation” vs imitation: The dispute centers on AI memorisation. Penguin Random House claims the model’s output shows it has retained and reproduced copyrighted material, indicating unauthorised use of training data. The company claims that AI companies such as OpenAI assert that their models do not store content but learn patterns, and that outputs are generated afresh rather than retrieved from stored sources.
An OpenAI spokesperson said the company is reviewing the allegations and reiterated its respect for creators and content owners, adding that it has been engaged in “productive conversations with many publishers around the world.”
“Human creativity is and remains at the heart of our work as publishers,” said Carina Mathern, the Penguin Random House Verlagsgruppe publisher for children’s and young-adult books. “We are first and foremost obliged to represent the interests of our authors and creatives.”
Mathern added that the company was “fundamentally open to the opportunities offered by AI, but at the same time, the protection of intellectual property is our top priority.”
Courts Recognise AI “Memorisation” as Copyright Infringement: A recent ruling by the Munich Regional Court in the GEMA vs OpenAI case provides important context for the Penguin Random House dispute by clarifying the legal risks of AI “memorisation.” The court determined that storing copyrighted material in AI models and reproducing it in outputs can both constitute copyright infringement, rejecting the argument that such use is incidental to training.
The court found that ChatGPT could reproduce song lyrics from its training data, even with simple prompts, showing that protected works can be embedded and later extracted from AI systems. As one of the first European decisions to hold an AI developer directly liable for both training and output, this ruling highlights the central issue in the Penguin Random House case: whether generative AI systems unlawfully retain and reproduce copyrighted content.
Why this case matters: The lawsuit may significantly influence how courts worldwide interpret AI-related copyright issues. It could set key precedents on the legality of training AI models with copyrighted data and clarify what constitutes infringement in AI-generated outputs. This is particularly relevant given previous German court rulings that found violations in certain AI training practices.
Read more:
For You
- Read Reasoned by Nikhil Pahwa: How AI is changing our world
- Sign up for MediaNama’s Daily Newsletter to receive regular updates
- Sponsor a MediaNama Event
The parliamentary panel found that the Finance Ministry is already using AI tools to detect financial fraud, including tax credit claims, fake invoices, circular trading, etc., but the ED is yet to use AI in their system.
MediaNama seeks transparency from MeitY on potential surveillance risks linked to ChatGPT and OpenAI tools, raising concerns over US laws, national security, and government guidelines on AI use.
MediaNama is the premier source of information and analysis on Technology Policy in India. More about MediaNama, and contact information, here.
© 2024 Mixed Bag Media Pvt. Ltd.
source