Search
Anthropic Enhances Claude Model with Termination Feature for Malicious User Queries
2025-08-18 /
Read about 0 minute
Author：小编
Anthropic has introduced a conversation termination feature in its Claude Opus 4 and 4.1 models. This feature activates when users persist in requesting the creation of harmful or abusive content, despite the model's repeated prior rejections of such demands. This initiative stems from Anthropic's research on the psychological strain experienced by AI models, with the goal of safeguarding the model's "well-being." As a measure of last resort, this feature allows the conversation to be restarted or the prompt to be revised following termination.
-
C114 Communication Network -
Communication Home
7 X 24 Track global technological trends
