Anthropic扩大Claude水印覆盖但留有参数漏洞
据The Decoder报道，Anthropic计划为Claude Mythos 5、Claude Opus 4.7、Claude Opus 4.6和Claude Sonnet 4.6加入文本水印，Claude Opus 5.5已配备该功能。在Opus 4.6和Sonnet 4.6上，仅当temperature、top_p和top_k保持默认值时才会应用水印，API用户修改任一参数即可规避。Anthropic称水印采用难以察觉的措辞模式，不增加token、不改变定价和API格式，也几乎不影响输出速度，但报道未提供独立测量结果。
此次更新让Anthropic能在多个托管平台上部署模型级水印以应对欧盟《人工智能法案》期限要求，但两款较旧Claude模型的覆盖仍取决于采样设置。
Anthropic's watermark rollout leaves a loophole on two older Claude models
Matthias Bastian View the LinkedIn Profile of Matthias Bastian
Oct 31, 2026
Nano Banana Pro prompted by THE DECODER
Anthropic is expanding its text watermark to four more Claude models, but API users can dodge the mark on two of them by tweaking a single sampling parameter.
#Anthropic#Claude Mythos 5#Claude Opus 4.7#Claude Opus 4.6#Claude Sonnet 4.6#Claude Opus 5.5#Amazon Bedrock#Amazon Web Services#Google Cloud#Microsoft Foundry#Google DeepMind#SynthID