HN 提问：AI 水印是否会带来新的攻击向量？
2 分•作者: nathanfig•大约 2 个月前
我从 Claude 关于水印的文档[0]中未能确定的是，他们会存储与特定水印关联的哪种元数据。表面上看，他们可以使指纹尽可能独特，甚至精确到具体的时间、用户和会话。
如果是这样，这似乎是一种隐藏的风险，AI 用户可能没有考虑到。你现在编写的任何代码都可能包含你不希望泄露的信息。如果恶意行为者获得了密钥，他们可能会揭露那些希望保持匿名的开源贡献者的身份。或者，如果足够多的指纹出现在一个样本中，可能会暴露公司不愿披露的内部组织细节。
更有想象力的人可能能想到更好的例子，这似乎是一个我没有看到太多考虑的攻击向量。
[0] https://support.claude.com/en/articles/16266773-how-claude-marks-ai-generated-content
查看原文
One thing I have not been able to determine from Claude's documentation on watermarking[0] is what kind of metadata they store in association with a given watermark. Ostensibly they could make the fingerprints as unique as they want, possibly down to the exact time, user and session.<p>If so, this seems like hidden risk that AI users are probably not considering. Any code you write now carries information that you might not want revealed. If a bad actor gets the keys then they may be able to de-anonymous open source contributors who want to stay hidden. Or perhaps enough fingerprints across a sample could reveal internal organization details a company would rather not disclose.<p>Someone with more imagination can probably come up with better examples, it just seems like an attack vector that I haven't seen much consideration for.<p>[0] https://support.claude.com/en/articles/16266773-how-claude-marks-ai-generated-content