Anthropic has agreed to settle a copyright infrigement lawsuit with authors to the tune of $1.5 billion US (with interest). The settlement still needs to be approved by the courts. Anthropic was accused of illegally using copyrighted works to train their (AI) LLM’s (large language models). Let’s be clear here; these companies (including Meta and Perplexity) are committing outright theft while both Anthropic and Meta have previously receiving favorable court rulings.
In the case of Anthropic, they didn’t just utilize downloaded online pirated material, they also purchased physical books, digitized (scanned) their pages, and then discarded all of that physical material without properly licensing/compensating the authors of those books (digital and physical).
This ruling (which again is not completely final yet) translates to authors being compensated around $3,000 per book with around 500,000 different pieces of work that would be eligible. Additionally, Anthropic agreed to destroy all data that had been collected for this LLM training.
This lawsuit and settlement should it stand, does at least set a strong precedent that favors the rights of the intellectual property rights holders. I still believe that Google, Meta, and OpenAI (the much larger companies that are misappropriating intellectual property) need to be held accountable for their own respective data training using pirated/non-licensed/non-compensated works.
