Key Highlights
- Solana co-founder Anatoly Yakovenko says AI companies should be allowed to train on data that people voluntarily publish online.
- Anthropic’s $1.5 billion copyright settlement highlights the ongoing legal debate over AI training and fair use.
- The court’s future decisions could shape how AI companies use data and whether blockchain-based data tracking becomes more important.
Solana co-founder Anatoly Yakovenko has weighed in on the legal debate surrounding artificial intelligence, arguing that AI companies should be allowed to learn from information people voluntarily publish online.
In a reply to an X post, Yakovenko said U.S. fair-use laws should protect AI companies such as Anthropic when they use publicly available online data to train their models. He wrote, “The judge ruled that training on books is fair use if they buy the books. So basically the opposite point of what you are trying to say.”
His argument is that if someone chooses to put information on the internet, AI companies should generally be able to use that information to learn. But the legal debate is not that simple.
Anthropic case puts AI training under the spotlight
Yakovenko’s comments come as U.S. courts continue to examine how copyright law applies to AI training. One of the clearest examples is Anthropic, the AI company he specifically referenced.
On Monday, US District Judge Araceli Martínez-Olguín approved a $1.5 billion settlement between Anthropic and a group of authors. The settlement ended the largest copyright class-action lawsuit ever certified and became the largest copyright settlement ever reached.
The case focused on Anthropic’s use of books to train its AI models. Earlier, the court ruled that training AI on books that were legally obtained could qualify as fair use. However, the use of pirated copies was likely not protected.
Authors challenge the settlement
That distinction is important because it shows why the debate around AI training data remains complicated. The question is not only whether AI companies use copyrighted material. Courts are also looking at how the material was obtained and how it was used.
Some authors challenged the settlement, arguing that the lawyers’ fees were too high while the payments to authors were too low. The proposed settlement was expected to pay around $3,000 for each work. Some authors wanted to reject the agreement and pursue separate lawsuits in the hope of receiving more serious damages.
How US fair use law comes into play
However, Judge Martínez-Olguín rejected the objections. She said about 95% of the class had received notice of the settlement, while around 91% of affected authors and publishers had already filed claims. Only 350 class members opted out, while another 54 people tried to object or submit late requests to leave the settlement.
The case is connected to Section 107 of the US Copyright Act, which sets out the rules for fair use. Courts consider factors such as why the material was used, the type of copyrighted work involved, and whether the use harms the market for the original work.
AI companies are now being tested against those rules, but courts have not reached one clear answer for every situation.
Yakovenko’s argument focuses specifically on information that people voluntarily publish online. That is a narrower position than the wider data practices used by some AI companies. His comments do not mean that every type of online data can automatically be used for AI training.
Why the outcome matters for AI and blockchain
The debate also has a connection to the growing relationship between blockchain and AI. Yakovenko did not announce a new Solana project, AI partnership, token launch, or product. However, the legal outcome could affect the need for blockchain systems that track where data comes from and who owns it.
If courts broadly protect AI companies that use publicly available data, systems for tracking data ownership and licensing may become less urgent. If courts impose stricter rules, AI companies could need stronger systems to track data origins, manage permissions, and compensate creators.
Also Read: TeraWulf Stock (WULF) Soars 15% as Anthropic Lease Fuels AI Pivot
