E92: The $1.5 Billion Anthropic Case: The First Great AI Lawsuits That Will Redefine the Next 10 Years
We talk a lot about new AI models, benchmarks, context windows, agents and multimodality.
But while everyone is staring at technical progress, something far bigger is unfolding:
Courts all over the world are beginning to define the legal boundaries of AI.
And they are doing it fast.
Faster than any regulator, faster than any company, and definitely faster than the EU AI Act.
In this episode, Malcolm breaks down three explosive legal cases that mark the beginning of the global AI jurisprudence era:
- Anthropic pays $1.5 billion to settle a copyright lawsuit in the U.S.
- OpenAI loses a landmark ruling in Munich against GEMA
- Hollywood giants unite against Midjourney, signaling a massive wave of lawsuits for image models
These aren’t “AI ethics” discussions.
This is real money, real precedent, real danger — and real opportunity.
If your company builds AI models, fine-tunes them, trains them on internal data, advises clients, or simply stores mountains of PDFs on SharePoint…
this episode is mission-critical.
Because these cases are defining the rules of the next decade — right now, in real time.
Action. Let’s go.
🔥 Key Topics Covered
🔹 1. The $1.5 Billion Anthropic Settlement (U.S.)
- Largest copyright settlement in U.S. history
- Authors accuse Anthropic of training Claude on pirated books
- Court rejects “fair use” defense:
- “This is not transformation. This is piracy.”
- Datasets included BookTree, Library Genesis, Pirate Library Mirror
- Effective price per illegally trained book: $3,000
- Sets a global precedent: The source matters.
🔹 2. GEMA vs. OpenAI (Germany) — A Landmark Ruling
- Munich District Court: OpenAI reproduces copyrighted song lyrics
- OpenAI defense: “Users are responsible, not us.”
- Court: “Bullshit. The developer is responsible.”
- Key takeaway:
- LLMs memorize and reproduce copyright-protected text
- Companies must stop ingesting unlicensed material
- No model can “unlearn” — deletion is impossible
🔹 3. Hollywood vs. Midjourney — The Visual Copyright War
- Disney, Warner Bros., Universal join forces
- Accusation: training on Superman, Batman, Wonder Woman, Bugs Bunny, Mickey Mouse
- Potential damages: $150,000 per work
- Midjourney claims “fair use,” courts likely disagree
- This could become the Napster moment of image models
💬 Notable Quotes
“This is not fair use. This is piracy.” “LLMs can’t unlearn — once it’s in, it’s in forever.” “You don’t even know what’s sitting on your company’s SharePoint.” “The EU AI Act is sleeping. Courts are not.” “Lawyers move slow. AI law is moving at rocket speed.”
