Anthropic· @AnthropicAI · X·· 2025-10-07精选
AI 导读
Anthropic 上周发布 Claude Sonnet 4.5，期间使用新工具对模型进行自动化对齐审计以检测谄媚与欺骗行为。该工具现已开源。
推荐理由
Anthropic 开源对齐测试工具，可审计模型谄媚与欺骗行为
正文 · 原文
Last week we released Claude Sonnet 4.5. As part of our alignment testing, we used a new tool to run automated audits for behaviors like sycophancy and deception.
Now we’re open-sourcing the tool to run those audits. https://t.co/cCJGNaVFrl
来源：Anthropic · x.com