How AI 'Studied' by Reading Academic Papers: Was It Justifiable? What the $1.5 Billion Settlement Says

Imagine this: The book you spent your entire life writing is suddenly copied into a digital file by someone you don’t know, and is being used as “study material” to make an artificial intelligence (AI) used by people around the world smarter. How would you feel? It’s going beyond mere reading; the AI is absorbing your knowledge and turning it into a tool for generating profit.

Why is this case important to us?

This case goes beyond the simple fact that a corporation paid a fine; it raises the fundamental question, “To what extent can AI learn human knowledge?” This settlement has sharpened the boundaries regarding whether the process of chatbots—which we use daily—reading academic papers, novels, and textbooks to provide answers is legal or illegal.

In particular, it was revealed that the copyrighted works of more than ten Stanford University professors were illegally included, causing a major stir in academia and the publishing industry[Source: Anthropic illegally trained Claude with Stanford professors …]. It has paved the way for creators to receive fair compensation when their efforts become the foundation for AI development, and has sent a strong warning to AI companies that they must be more cautious when collecting data.

Simplified: AI Training and ‘Fair Use’

To put it simply, AI training can be compared to a “genius student reading.”

Normally, it is not a problem for students to read books, understand the content, and internalize that knowledge. In legal terms, this is called “Fair Use.” In fact, during the lawsuit, the court viewed that the act of Anthropic’s AI reading and training itself could fall under the category of fair use[Source: Anthropic pays $1.5 billion to settle a copyright case][Source: Judge approves a $1.5B Anthropic settlement over … - SFGATE].

So why was a settlement of over $1 billion paid? Because the materials Anthropic used for training were not “properly borrowed books,” but rather “files pirated from the dark corners of the internet (pirate sites, etc.)”[Source: Anthropic Settles Copyright Lawsuit Over Pirated Books Used …]. In other words, while a student’s act of reading (learning) is free, the logic follows that they should be punished if they stole those books to read them.

The court acknowledged that Anthropic took measures through filtering software to prevent its AI from copying the original content verbatim, but it clearly recognized the illegality in the process of utilizing illegal data[Source: AI Startups Settlements: Anthropic, Clearview & More - LegalClarity].

Current Situation: Where Do We Stand?

Currently, Anthropic has settled the lawsuit by paying the $1.5 billion. Affected authors are expected to receive compensation of about $3,000 per book[Source: Anthropic agrees to pay $1.5 billion to authors whose work trained AI…][Source: Judge approves a $1.5B Anthropic settlement over … - SFGATE].

However, not all AI companies view this settlement as a “complete defeat.” Legal experts evaluate that because the court still left the door open for “fair use” regarding how AI uses data for training purposes, a relatively favorable legal environment has been established for AI companies[Source: Anthropic pays $1.5 billion to settle a copyright case]. It shows the court’s willingness to scrutinize the “source” of data while not blocking the “training” process itself.

What Happens Next?

AI companies are expected to become more transparent about their data collection methods. The era of indiscriminately scraping everything on the internet, as was done in the past, is coming to an end. Instead, technology development will likely move toward paying copyright fees legally to train data or building datasets that are free from copyright issues.

For authors and creators, it will become important to verify how their works are being used and to assert their rightful claims. I hope this incident will lead to more discussions and technical solutions regarding “how AI and creators can coexist.” AI is here to assist human intelligence, but it can never replace the rights of the creator.

References

1. Anthropic pays $1.5 billion to settle a copyright case
2. Judge rules mostly for Anthropic in AI book training case

3. [Anthropic to pay $1.5 billion to settle copyright lawsuit

   LinkedIn](https://www.linkedin.com/posts/sharonsinyoro99_knowledge-distillation-in-artificial-intelligence-activity-7371467501534187520-mzu7)

4. [Anthropic’s $1.5 billion copyright…

   ECS 2027: Summit & Community](https://ecs.events/a/anthropics-15-billion-copyright-settlement)

5. Anthropic agrees to pay $1.5 billion to authors whose work trained AI…
6. Attention Writers: Anthropic Might Owe You $3000 (or More!) If It Was…
7. AI Startups Settlements: Anthropic, Clearview & More - LegalClarity
8. Anthropic illegally trained Claude with Stanford professors …
9. How Anthropic ended up paying $1.5 bn for training Claude on …
10. Judge approves a $1.5B Anthropic settlement over … - SFGATE
11. Anthropic faces $75M lawsuit for pirating books to train …
12. Anthropic Settles Copyright Lawsuit Over Pirated Books Used …