The judge reigns the formation of AI on documents protected by anthropic on the material protected by copyright

Anthropic has A mixed result in a collective appeal brought by a group of authors who said that the company had used its creations protected by copyright without authorization. On the positive side of the artificial intelligence company, the senior district judge William Alsup of the American district court for the North District of California determined that the training of anthropic on its AI tools on works protected by copyright was protected as a fair use.

The development of models of large languages ​​for artificial intelligence has created a boondoggy of copyright law while creators try to protect their works and their technological companies Or To bring together more training equipment. Alsup’s decision is one of the first ones that will probably establish the basics of legal precedents around what AI tools can and cannot do.

The use of copyright equipment can be considered equitable use if the exit is determined as “transformer” or not a substitute for the original work. “The technology in question was among the most transformative that many of us will see in our life,” said Alsup .

Despite the equitable designation of use, the decision always provides some appeal to writers; They can choose to bring anthropic to justice for hacking. “Anthropic has downloaded more than seven million pirated copies of books, did not pay anything and kept these pirated copies in her library even after having decided that she would not use them to form her AI (all or never),” wrote Alsup. “The authors argue that Anthropic should have paid for these pirated library copies. This order agrees.”