Humanity Settles Class Action Lawsuit for $1.5 Billion
A federal judge learned on Friday that Humanity has agreed to pay $1.5 billion to settle a class action lawsuit brought by a group of authors. They accused the AI company of using their books to train the AI chatbot Claude without obtaining proper permissions.
The plaintiffs, represented in court, asked District Judge William Alsup to approve this settlement, although specific terms and details were not disclosed at that time.
The plaintiffs’ filing stated, “If approved, this landmark settlement will represent the largest publicly reported restoration of copyright in history, surpassing previous copyright class action cases.”
This proposed settlement is notably the first in a series of lawsuits against major tech companies, including OpenAI, Microsoft, and Meta, focused on how copyrighted materials are utilized to train generative AI systems.
As part of the agreement, Humanity pledged to destroy its copies of the books the authors claimed were pirated. Additionally, the settlement suggests that the company could still face infringement claims regarding material generated by its AI.
In a statement, Humanity emphasized its commitment to developing safe AI systems aimed at enhancing people’s capabilities and solving intricate problems. It’s worth noting that this contract does not include any acknowledgement of liability.
Authors Andrea Burtz, Charles Graeber, and Kirk Wallace Johnson filed the class action lawsuit against Humanity last year, alleging that the company, which is backed by Amazon and Alphabet, had illegally used millions of pirated copies to train the AI assistant Claude.
The writers’ claims resonate with many other lawsuits asserting that tech companies have exploited creative works from various authors, news outlets, and visual artists for AI training.
In defense, companies contend that their systems use copyrighted material in a fair manner to create new and transformative content.
Earlier in June, Judge Alsup decided that while Humanity had used the authors’ work fairly to train Claude, the company had infringed rights by storing over 7 million unauthorized copies in a “central library,” which may not have been intended for that purpose.
The trial is set to start in December, focusing on suspected copyright infringement, with potential damages that could soar into the hundreds of millions.
This case stirs significant fair use questions, which are actively being explored in parallel AI copyright cases.
Another San Francisco judge is currently overseeing a similar case against Meta, which surfaced shortly after Alsup’s ruling, noting that utilizing copyrighted work without permission for AI training is illegal in various contexts.
