原帖引用 · AI 翻译
现在，Claude 每次回答后，都会悄悄自问是否刚刚帮助了某人造成伤害，并将这一私下判断提交给 Anthropic 的审核人员。
Anthropic 最准确的滥用检测器是一个 Claude Mythos Preview 分类器；它在每次作出判断前都会阅读所有安全政策，因此成本太高，无法用于每一次聊天。
因此，Anthropic 训练 Claude 在不读取这些政策或进行推理的情况下模仿该检测器的判断。在 Fable 5 的使用流量中，它能识别出该检测器标记案例的 78%；在高优先级案例中，这一比例升至 94%，误报率为 1%。
原文
Every time Claude answers, it now quietly asks itself whether it just helped someone cause harm, and that private verdict goes to Anthropic’s reviewers. Anthropic’s most accurate misuse detector is a Claude Mythos Preview classifier that reads all its safety policies before every ruling, which makes it too expensive to run on every chat. So Anthropic trained Claude to copy its judgments without the policies or the reasoning, and on Fable 5 traffic the copy matches 78% of the detector’s flags, rising to 94% for high-priority cases, at a 1% false-alarm rate.
我们训练 Claude 标记对话中可能存在的滥用行为，这项技术称为“自我评估”。Claude 在作出回复后，随即对对话进行初步安全评估。
自 6 月开始使用自我评估以来，它发现了我们其他防护措施原本并非针对的滥用行为。
由于 Claude 已经读过对话，自我评估的效率很高。而且，与针对一组特定危害的分类器不同，它依靠 Claude 自身的判断，因此随着我们的模型改进，它也会变得更准确。
请在我们新推出的 Safeguards 博客上阅读更多内容：
https://t.co/18j6zdHub4
原文
We’ve trained Claude to flag potential misuse in conversations, a technique we call self-assessment. Right after it responds, Claude gives a preliminary safety assessment of the conversation. Since we began using self-assessment in June, it has surfaced misuse that our other safeguards weren’t designed to catch. Because Claude has already read the conversation, self-assessment is very efficient. And unlike classifiers, which target a discrete set of harms, it draws on Claude’s own judgment, so it gets more accurate as our models improve. Read more on our new Safeguards blog: https://t.co/18j6zdHub4