GEMA, the collecting society, prevailed in its copyright lawsuit against OpenAI in Germany.
According to Reuters, which first reported the decision, “both the memorisation in the language models and the reproduction of the song lyrics in the chatbot’s outputs constitute infringements of copyright exploitation rights.”
“Presiding judge Elke Schwager ordered OpenAI to pay damages for the use of copyrighted material, without disclosing a figure.”
The opinion in German can be downloaded here (hat tip: ai fray).
Is this a harbinger of decisions to come?
With many copyright lawsuits against OpenAI around the world, one can’t help but wonder if today’s ruling is a harbinger of decisions to come for OpenAI?
Of course, the fair use defense is distinctive to the U.S. law — it provides a different ground for noninfringement (based on a general, case-by-case exception versus a limited specific exception) than available in other countries, including Germany.
But OpenAI also faces copyright suits outside of the United States, in Canada, Brazil, and India. Canada and India recognize fair dealing, which is more circumscribed. And Brazil is a civil law system similar to Germany.
And even in the United States, OpenAI faces a daunting challenge of defending against roughly 13 copyright lawsuits, a bunch proposing class actions for book authors. They all are before Judge Stein, sitting as a court in MDL jurisdiction. At least for summary judgment, OpenAI is at the mercy of the MDL court for all 13 lawsuits.
Even putting aside the lawsuits filed by the New York Times and other newspapers, the Class suits against OpenAI allege the separate theory of infringement (apart from training) based on OpenAI’s alleged acquisition and use of copies from shadow libraries.
This so-called Shadow Library Strategy is what prompted Anthropic to settle the class action lawsuit against it, similarly filed by book author, for a historic $1.5 billion.