Big Win for AI, Bigger Win for Authors: What the Landmark Anthropic Court Ruling Means

We're seeing AI copyright cases come before the court system and sorting through their rulings. These cases will have direct ramifications on our industry and on how we're allowed to use AI moving forward.

A recent federal court ruling has provided a nuanced outcome in the fair use lawsuit brought by authors and studios against Amazon-backed AI company, Anthropic.

While the decision marks a key victory for Anthropic regarding the use of lawfully acquired copyrighted material for AI training, the company still faces an impending trial over allegations of utilizing pirated books.

Let's break it down.

Big Wins for AI

As reported in The Hollywood Reporter, the lawsuit, filed by authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson, accused Anthropic of large-scale copyright infringement, claiming the company profited from their works without authorization by using them to train its AI model, Claude.

Obviously, this is something we should care about, because we know that AI companies are scraping the internet and gathering all the information, then using that information to create their own posts.

If that information was copyrighted, this would represent an infringement on that.

U.S. District Judge William Alsup's ruling found that Anthropic's use of legally purchased books to train its large language models (LLMs) was "quintessentially transformative" and therefore constituted "fair use" under U.S. copyright law.

In layman's terms, it's saying that if the AI was trained on books they paid for, then that's basically akin to learning, and you can't call that ripping the books off if it uses information it "learned" from them.

This is seen as a crucial precedent for AI development, suggesting that training AI on lawfully obtained copyrighted works may not infringe on copyright.

The judge said that, like a human reader, Anthropic's LLMs learned from the works to create something new and different, rather than simply replicating or supplanting the originals.

Bigger Wins for Authors

There's one key aspect to all this that I do think will affect how AI companies move forward.

Anthropic's alleged creation and maintenance of a "central library" of millions of pirated books was deemed to violate copyright and not fall under fair use.

This means Anthropic will face a separate trial in December to determine potential damages for its use of these illicitly obtained materials.

That's a huge development because it means AI companies cannot just scrape the internet and take things; they have to pay for them.

Court documents also revealed internal concerns among Anthropic employees regarding the legality of using pirate sites to access books, with the company later shifting to purchasing and scanning physical books.

They were right to be concerned; it was against the law.

Where We Go From Here

This mixed ruling has generated varied reactions. Some view it as a victory for technological innovation, potentially accelerating AI advancements by validating the use of existing data.

Others, particularly content creators, express concerns that, while AI training on legally acquired material may be permissible, the issue of pirated content remains a significant threat to intellectual property rights and livelihoods.

And it proves that at these companies, despite people's concerns internally, there's theft going on in terms of what they can get from the internet.

And Anthropic isn't the only one facing litigation — Midjourney is facing a similar lawsuit against Disney, Universal, and other major Hollywood studios.

Summing It All Up

The Anthropic case serves as a pivotal moment, shaping how AI companies approach data acquisition and usage, and underscoring the ongoing need for clarity and consistency in legal standards governing AI development.

The outcome of the December trial will further define the legal boundaries for AI in the realm of copyright.

And I'm rooting for the humans.

Let me know what you think in the comments.