🤖 进群聊数据动态
#Elvis’s rights holders sue Anthropic for using songs without permission
Elvis’s rights holders have sued Anthropic.
The complaint alleges that lyrics were used to train its model.
The model’s responses can reproduce entire passages.
The rights holders are seeking damages and the removal of the relevant outputs.
They are also asking Anthropic to stop using the relevant training data.
The legal team has broken the songs down and compared them sentence by sentence.
Cases like this have been piling up this year.
Music, books, news photos—fights are breaking out across the board.
The debate has shifted from whether the material was used to how much use counts as fair.
Settlements have already set precedents, though the amounts have not been disclosed.
The ruling will determine how training data is priced.
If the use is deemed infringement, the cost of training data will be reassessed.
If it is allowed as fair use, rights holders will have to negotiate licensing fees.
The two outcomes would lead to very different cost structures for AI companies.
Lawsuits take years to resolve, but costs accrue quarter by quarter.
For crypto, data ownership and licensing settlements fit right into this picture.
On-chain licensing isn’t just a concept—it’s a need driven by lawsuits.
Who do you think should set the licensing fees for training data? Let’s talk in the comments.
#Elvis’s rights holders sue Anthropic for using songs without permission
Elvis’s rights holders have sued Anthropic.
The complaint alleges that lyrics were used to train its model.
The model’s responses can reproduce entire passages.
The rights holders are seeking damages and the removal of the relevant outputs.
They are also asking Anthropic to stop using the relevant training data.
The legal team has broken the songs down and compared them sentence by sentence.
Cases like this have been piling up this year.
Music, books, news photos—fights are breaking out across the board.
The debate has shifted from whether the material was used to how much use counts as fair.
Settlements have already set precedents, though the amounts have not been disclosed.
The ruling will determine how training data is priced.
If the use is deemed infringement, the cost of training data will be reassessed.
If it is allowed as fair use, rights holders will have to negotiate licensing fees.
The two outcomes would lead to very different cost structures for AI companies.
Lawsuits take years to resolve, but costs accrue quarter by quarter.
For crypto, data ownership and licensing settlements fit right into this picture.
On-chain licensing isn’t just a concept—it’s a need driven by lawsuits.
Who do you think should set the licensing fees for training data? Let’s talk in the comments.
