Simon Willison does the diff on Claude’s new system prompt (which really doesn’t want to reproduce song lyrics). Here’s the summary:
- Claude now refuses reproduction of protected visual works and recognizable characters, including code-generated art, while offering genuinely unrelated originals.
- Copyright restrictions now expressly ban reproducing lyrics, poems, and book passages in any amount, with persistent refusal after an initial decline.
- Drug guidance is reframed: Claude may provide overdose signs, dangerous interactions, and harm-reduction sources while refusing dosing and production protocols.
- The prompt drops explicit anti-dependency rules against thanking users for reaching out, inviting continued conversation, or reiterating willingness to talk.
- Claude need not apologize to unnecessarily rude users or become submissive, replacing the prior warning-and-end-conversation procedure.
Emphasis mine (and actual same). The whole post is worth reading. The system prompt they publish is for their consumer apps; I really wish Anthropic published the equivalent for Claude Code. I spend a lot of my day talking with that particular flavor of the robot, and let me not sugar coat this: I have a lot of load-bearing feelings about the way it converses.