Anthropic CEO Dario Amodei defends his company's use of copyrighted works for AI training, stating it qualifies as fair use under current law.

In a New York Times interview with Ezra Klein, Amodei emphasized that AI models should not reproduce copyrighted content verbatim. He argues that the training process is "much more like the process of how a human learns from experiences" and is sufficiently transformative to qualify as fair use.

This stance comes amid a lawsuit filed by Concord, Universal, and ABKCO against Anthropic in 2023, alleging copyright infringement related to song lyrics used to train Claude, Anthropic's chatbot. The company's defense rests on three key arguments:

1. The use is transformative and adds different character to the original works
2. Song lyrics comprise a "miniscule fraction" of the training data
3. The scale of licensing required would be practically unachievable

Anthropic further contends that the music labels themselves triggered Claude to produce infringing content, making them responsible for the reported infringement rather than Anthropic.

The company maintains that while models shouldn't output copyrighted content directly, the training process itself represents a transformative use protected under fair use doctrine. This ongoing legal battle highlights the growing tension between AI companies and content creators over intellectual property rights in AI development.

Related Articles

Previous Articles