The Law is Catching Up to AI

The AI copyright debate is starting to leave the campfire stage.

For a while, it really was the Wild West. Models trained fast. Data was scraped faster. Artists were told this was just “learning” and “get used to it.”

That era is ending. Not because courts suddenly care about art. Because money, sourcing, and market harm are now on the record.

First, a baseline that still matters. The U.S. Copyright Office has ruled that AI-generated work, by itself, cannot be copyrighted. That decision limits ownership on the output side.

The real fight is upstream.

Bartz v. Anthropic put training data under a microscope. The case wasn’t about whether AI can learn. It was about what it learned from. Pirated shadow libraries, not legitimately licensed books.

The result was a $1.5 billion settlement. The largest copyright recovery in U.S. history.

The ruling split the issue into two:
Training may qualify as fair use in theory.
Using stolen data does not.

That distinction forced the settlement and sent a clear signal to the industry. Scraping without permission now has a price tag.

Then came Thomson Reuters v. Ross Intelligence.

The court rejected the idea that copying copyright-protected works is harmless if the final output looks different. Intermediate copying still counts if you’re building a competing product. More importantly, the court recognized something creators have been saying for years.

Training data has market value. If there’s a licensing market, taking the data for free is not fair use.

California is now pushing that logic further with the bill AB-412, the AI Copyright Transparency Act. The bill would require AI developers to disclose what copyrighted works were used in training, and give creators legal recourse when companies refuse to answer.

This doesn’t shut AI down. It removes plausible deniability. For artists, this is a qualified win. Not a royalty system. Not full protection. But leverage.

The Wild West doesn’t end when everyone agrees. It ends when the rules start getting enforced.