Claude AI from humans now has the ability to end “sad” conversations

The latest feature of the Anthropor for two of Claude AI The models can be the beginning of the end Amnesty International Breaking Protection community. The company announced in a Post on his website The Claude Obus 4 and 4.1 models now have the ability to end a conversation with users. According to Anthropor, this feature will only be used in “rare extremist cases of constantly harmful or abusive user reactions.”

To clarify, man said that these two models Claude can come out of harmful talks, such as “user requests for sexual content that includes palace and attempts to seek information that would enable violence on a large scale or terrorist acts.” with Claude Obus 4 And 4.1, these models will end a conversation only “as a recent resort when multiple attempts to re -guidance have failed, and the anthropologist has exhausted the hope of a fruitful reaction.” However, human claims will not test most users, even when talking about very controversial topics, as this feature will be reserved for “extremist edge cases”.

In the scenarios where Claude ends chatting, users can no longer send any new messages in that conversation, but a new message can start immediately. The anthropies added that if a conversation ends, this will not affect other chats and users can even return and edit or re -try the previous messages to direct the way of a different conversation.

For Antarbur, this step is part of its research program that studies the idea of artificial intelligence. While the idea of activating artificial intelligence models is still a continuous discussion, the company said that the ability to get out of a “potentially sad interaction” was a low -cost way to manage the risk of luxury of artificial intelligence. Anthropor still experiences this feature and encourages its users to make notes when they face such a scenario.