Hello All-
We at Brick Cave wanted to comment on the recent proposed settlement between AI startup Anthropic and the Author community in regard to Anthropic inappropriately scraping books to train its large language models against author permission.
Reuters has a good article here summarizing the status of the litigation to this point. There are additional articles at Wired or any number of outlets for you to gather perspective on this issue. Author's Guild has a comprehensive article if you are an author and suspect your work is involved.
For full disclosure: Brick Cave authors and works have been found to have been scraped by Anthropic in this ongoing circumstance. The same is true of additional large language models that are continuing to perform this function against Brick Cave's wishes.
There are two key elements to this case:
The first, is that the judge has found that Anthropic's creation of "shadow libraries", collecting works without paying for them, keeping them regardless of the intent to use them for training their AI models, to be illegal. In essence, the settlement is centered around this finding. While Brick Cave agrees with the sentiment of the decision here, we are not sure that allowing Anthropic to simply provide a financial restitution is the best solution to the issue, as it now sets a price tag for the next violator to better plan with.
The second, far more controversial and troubling decision, is that Anthropic's use of these same books to train its AI models is allowed under the principle of "fair use" doctrine of copyright law.
As you might expect, as a book publisher, Brick Cave is not in alignment with this view. Anthropic's intent is to build a profitable AI business model using content that it had not licensed. Not just purchased, but actually gained permission to use the content in that manner. The vast majority of authors write a book to educate people, to entertain people, to engage people. Fair use is not enabling profit streams for unintended parties.
Ultimately, this represents continued entitlement by technology companies that they have unfettered access to well, everything, simply because they are technology companies and promising a "greater good". When, in fact, they are simply creating profit for themselves without really providing a material benefit to culture or those that created said content. In some cases, this technological entitlement creates new cultural challenges and problems.
To be clear, Brick Cave is not "Anti AI". The roots of AI were deeply imbedded in the publishing industry the moment the first computer was engaged in the industry. We are not naive to the benefits that the progress in AI can provide to our business, our industry and our community. We DO believe that those benefits should not come at the expense of the rights of the individual creative to have agency in the use of their work, or at the very least receive compensation appropriate to it's use.
Anthropic anticipates that it will generate BILLIONS in revenue from the creation of their AI models. This settlement amount in that context amounts to a parking ticket. Yet, the knowledge their models will have gained from the works they used will drive their profit for the entire time of Anthropic's existence.
