Judge Says Anthrophic Can Steal Your IP

This is a critical issue for the AI industry and this opinion was issued by a district court judge, so don’t count those chickens or eggs or whichever comes first just yet.

The district court judge is William Alsup. He is well known and has been on the bench for more than 20 years. He was appointed by Bill Clinton. Among the many well known cases he presided over were the Oracle v. Google Java case and the Waymo/Otto/Uber theft of IP case, so he certainly has the creds.

There are two pieces to this case. The first one, which for Anthropic, was that using books from authors to train its AI was fair use. The theory behind this is that the training is “transformative” meaning you are not just using the copyrighted material directly, but rather transforming it into something else; in this case an LLM.

The second part of the case didn’t go so well for them.

Anthrophic had copied and stored more than 7 million pirated books as part of their training process. That, he said, was not fair use.

Willful copyright infringement can draw fines of up to $150,000 per work.

$150,000 x 7,000,000 books is about a trillion dollars. Obviously, he is not going to fine them that much, but the fine could be significant.

This is not the only lawsuit being brought by authors, news sources and other copyright owners, so this will ultimately be decided by the supremes.

Anthropic’s defense, basically, is that it would be expensive to pay authors for their work, which would hamstring them. They convinced Alsup.

Anthropic and other prominent AI companies including OpenAI and Meta Platforms have been accused of downloading pirated digital copies of millions of books to train their systems. If this part of the decision holds, that could get expensive.

It will be years before this is settled, but it will wind its way through the courts, slowly.

Credit: Cybernews

by