According to CoinMeta, Anthropic announced that its AI model Claude will have the ability to actively terminate conversations, marking a shift in the way large language models interact with users. The new policy will take effect on November 12, 2026, and explicitly prohibits "continuous and unnecessary abuse or cruel treatment" of the models. The introduction of this mechanism means that AI is no longer just a passive service provider; it can now actively enforce behavioral norms, changing the power dynamics of human-computer collaboration.