Anthropic says some Claude models can now end ‘harmful or abusive’ conversations 

Anthropic says new capabilities enable its newest AI fashions to guard themselves by ending abusive conversations.