This week brought further examples of the two approaches content copyright holders are taking in their dealings with the AI platforms.
One is to take them to court, the other is to cut deals.
A group of authors filed a class action against Anthropic, accusing the company of using their copyrighted work to train their AI-powered chatbot, Claude.
According to the filing, “it is no exaggeration to say that Anthropic’s model seeks to profit from strip-mining the human expression and ingenuity behind each one of those works.”
As reported by The Guardian, the authors said in their complaint that Anthropic had “built a multibillion-dollar business by stealing hundreds of thousands of copyrighted books.”
Late last year, the New York Times brought a similar action against OpenAI, accusing them of seeking to “free-ride on the Times’s massive investment in its journalism by using it to build substitutive products without permission or payment.”
Other copyright holders are taking a different approach. This week, Condé Nast followed the likes of Axel Springer, Le Monde and NewsCorp in cutting a deal with OpenAI.
The deal gives OpenAI permission to access the publisher’s content to both meet its ongoing need for training data and also for up-to-date and reliable information which will be a key component of its still-in-the-pipeline SearchGPT offering.
Condé CEO Roger Lynch told staff, “it’s crucial that we meet audiences where they are and embrace new technologies while also ensuring proper attribution and compensation for use of our intellectual property.”
Typically, financial terms were not disclosed, so we can’t tell exactly how "proper" the compensation is. Non-disclosure seems to be feature of these deals. I can’t imagine why...
At one level, the deal represents an acceptance that LLMs are here to stay, so better to cut a deal early because the financial value to publishers of such deals is likely to come down as more publishers sign up.
(Incidentally, Claude was used in the workflow of this article, but not, I hasten to add, in its creation...)
You can catch James Evelegh’s regular column in the InPubWeekly newsletter, which you can register to receive here.