The Ongoing Battle over Generative AI and Copyright
The UMG v. Anthropic Case: A Unique Exception
The one exception to the general trend of generative AI companies being sued for copyright infringement is the UMG v. Anthropic case. In this case, early versions of Anthropic’s AI generated song lyrics for songs, which posed a problem. However, the parties have since agreed to put safeguards in place to prevent this from happening, and the case is pending resolution.
The Hard Question for AI Companies
At the end of the day, the harder question for AI companies is not whether it is legal to engage in training, but rather what do you do when your AI generates output that is too similar to a particular work?
Settlements on the Horizon?
There may well be some settlements. Large players with significant content or valuable content may end up with settlements and licensing deals. For example, OpenAI may pay The New York Times to use their content.
Judgments and Class-Action Lawsuits
There will be significant judgments set by courts, and class-action lawsuits are likely. The defendants will likely resist these cases and aim to win on summary judgment. The Supreme Court’s decision in Google v. Oracle has nudged fair-use law in the direction of being resolved on summary judgment, not in front of a jury. AI companies will try to get these cases decided on summary judgment.
Why Summary Judgment?
It’s quicker and cheaper than going to trial. AI companies are worried that they won’t be viewed as popular if they’re found to have made a copy of the work that should be illegal, and people won’t dig into the details of the fair-use doctrine.
Deals Between AI Companies and Media Outlets
There have been many deals between AI companies and media outlets, content providers, and other rights holders. Most of these deals are about search, not foundational models. This is how it’s been described to me.
Licensing Content for AI Search Engines
If you’re using retrieval-augmented generation on targeted, specific content, your fair-use argument gets more challenging. The risky area is that AI-generated search is more likely to generate text taken directly from one particular source in the output, and that’s much less likely to be a fair use.
What You Need to Know
The thing I hear most often that’s wrong as a technical matter is that these generative AI systems are just plagiarism machines. They’re not. They’re generating new content that nobody could expect or predict. It’s giving us a lot of new content. I think that’s important and valuable.
Conclusion
The battle over generative AI and copyright is ongoing, with many cases pending and some already resolved. As the technology continues to evolve, it’s essential to understand the legal implications of its use.
FAQs
Q: What is the UMG v. Anthropic case about?
A: The UMG v. Anthropic case is a unique exception to the general trend of generative AI companies being sued for copyright infringement.
Q: What is the harder question for AI companies?
A: The harder question for AI companies is what do you do when your AI generates output that is too similar to a particular work?
Q: Will there be settlements on the horizon?
A: Yes, there may be settlements, especially with large players with significant content or valuable content.
Q: Why do AI companies want to win on summary judgment?
A: It’s quicker and cheaper than going to trial, and AI companies are worried about being viewed as popular if they’re found to have made a copy of the work that should be illegal.
