Skip to content

500 Songs at Risk: AI Chatbot Guardrails Strengthened in Copyright Clash

Anthropic, the company behind the Claude chatbot, has agreed to maintain stringent guardrails on its AI models to prevent the output of copyrighted song lyrics. This decision comes as part of a legal agreement with music publishers who claim that the chatbot has been reproducing lyrics from hits like Beyonce’s “Halo” and Bob Dylan’s “Like a Rolling Stone” without proper licensing. The agreement, overseen by US District Judge Eumi Lee, does not admit any wrongdoing by Anthropic but ensures that the chatbot’s outputs do not include lyrics from the 500 songs at the heart of the lawsuit. Additionally, Anthropic has committed to applying these guardrails to any new products or offerings. If the publishers find any infringing outputs, they can notify Anthropic, who will then provide a detailed response or explain their intent not to address the issue. This deal, while not resolving the broader copyright training issue, introduces a level of accountability into the AI’s operation.

Source: arstechnica.com

Related Links

Related Videos