AI & Me: If AI helped write it, should it be labelled?

Anthropic plans to embed invisible, machine-readable watermarks in text generated by Claude. But a watermark can only show that AI was involved, it cannot tell us whether Claude wrote the entire piece, heavily edited it or simply corrected the grammar. On AI & Me, Lynlee Foo asks where AI assistance ends and authorship begins, and whether disclosure should depend on what the AI actually did.

See omnystudio.com/listener for privacy information.