Claude AI of Anthropic now has the ability to end “painful” conversations

Anthropic’s latest feature for two of his Claude AI Models could be the start of the end for the Jailbreake community. The company has announced in a Publish on your website That the Claude Opus 4 and 4.1 models now have the power to end a conversation with users. According to Anthropic, this characteristic will only be used in “rare and extreme cases of constantly harmful or abusive user interactions”.

To clarify, Anthropic said that these two Claude models could leave harmful conversations, such as “requests from sexual content users involving minors and try to request information that would allow large -scale violence or acts of terror”. With Firm work 4 And 4.1, these models will put an end to a conversation “as a last resort when several attempts at redirection have failed and the hope of a productive interaction has been exhausted,” according to Anthropic. However, Anthropic claims that most users do not contain Claude to run a short conversation, even when talking about very controversial subjects, because this feature will be reserved for “extreme edges”.

In the scenarios where Claude ends a conversation, users can no longer send new messages to this conversation, but can create a new one immediately. Anthropic added that if a conversation is over, this will not affect other cats and that users can even go back and modify or try the previous messages to head to a different conversational route.

For Anthropic, this decision is part of its research program which studies the idea of AI well-being. Although the idea of anthropomorphizing AI models remains a continuous debate, the company said that the ability to leave a “potentially painful interaction” was a low cost means of managing the risks for AI well-being. Anthropic always experiences this feature and encourages its users to provide comments when they meet such a scenario.