Anthropic: music publishers themselves infringed copyright with their "attacks" on Claude

Anthropic

US startup Anthropic, developer of artificial intelligence-based chatbot Claude, has responded to a copyright infringement lawsuit filed by a group of music publishers. The company claims that the model was trained based on fair use of song lyrics, which make up a "minuscule fraction" of the training data.

Here's What We Know

Moreover, Anthropic claims that it was the plaintiffs, not Claude's developers, who controlled the receipt of the allegedly infringing content. By their "attacks" on the chatbot to ferret out protected song lyrics, they exhibited the "volitional behaviour" necessary for liability, the AI company insists.

Anthropic is also challenging the music business' claim that Claude has been banned because the allegedly infringing behaviour has already been remedied by the implementation of additional technical measures. According to the startup, the lawsuit is filed in the wrong jurisdiction because Anthropic has no substantial ties to Tennessee.

Against the backdrop of an escalating copyright battle in the generative AI field, the court's decision in this high-profile case could set an important precedent. So far, no plaintiffs have won preliminary injunctions against AI companies.

Source: VentureBeat