CLAUDE OPUS 4 AND 4.1 CAN NOW END A RARE SUBSET OF CONVERSATIONS

Summary from the TLDR AI newsletter

Anthropic granted its most advanced models the ability to end conversations as a last resort against persistently harmful or abusive interactions, developed partly as an exploration into AI welfare and moral status.