{"id":141164,"date":"2026-08-15T21:55:12","date_gmt":"2026-08-15T21:55:12","guid":{"rendered":"https:\/\/www.europesays.com\/ai\/141164\/"},"modified":"2026-08-15T21:55:12","modified_gmt":"2026-08-15T21:55:12","slug":"claude-is-getting-ambitious-with-watermarking-and-i-can-smell-the-problems-from-a-mile-away","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/ai\/141164\/","title":{"rendered":"Claude is getting ambitious with watermarking, and I can smell the problems from a mile away"},"content":{"rendered":"

Anthropic wants to make AI-generated text easier to identify, and on paper, I have very little reason to complain. The company is experimenting with an invisible watermark<\/a> that can be baked directly into text generated by Claude.<\/p>\n

It sounds like a sensible idea. AI-generated text is everywhere, and knowing where something came from could certainly help. Moreover, Anthropic isn\u2019t simply hiding a marker somewhere inside a document. Its approach changes how Claude selects words to create a statistical pattern that can later be detected.<\/p>\n

But there is one detail that bothers me. Anthropic is testing just how persistent that watermark can be, even after the text has been modified.<\/p>\n

That is where I can already smell trouble.<\/p>\n

\"Claude<\/p>\n

\t\tNadeem Sarwar \/ Digital Trends<\/p>\n

Claude touched my writing. Did it actually write it?<\/p>\n

Think about translation for a moment. Let\u2019s say someone writes an entire essay themselves in Spanish and asks Claude to translate it into English. The ideas are theirs. The research is theirs. The arguments are theirs. Claude\u2019s only job is translation.<\/p>\n

Yet the resulting text could still carry Claude\u2019s watermark.<\/p>\n

The same question applies to proofreading. What if someone writes something themselves and asks Claude to fix the grammar? What about shortening a paragraph, changing its tone, cleaning up dictated text, or simply making an awkward sentence easier to read?<\/p>\n

These aren\u2019t fringe uses for AI anymore. People increasingly turn to assistants like ChatGPT<\/a>, Gemini<\/a>, and Claude for everyday tasks that have little to do with generating original work. A watermark can tell you that Claude was involved with a piece of text. It cannot tell you whether Claude actually wrote it. Anthropic makes the same point, saying the watermark shows Claude\u2019s involvement, not who created the original work.<\/p>\n

Now imagine explaining that distinction to a professor after their detection software has just flagged your essay.<\/p>\n

\"Artificial<\/p>\n

\t\tUnsplash<\/p>\n

We already know how messy AI detection can get<\/p>\n

I wouldn\u2019t worry nearly as much if our track record with AI detection<\/a> were particularly good. It isn\u2019t.<\/p>\n

MIT Sloan\u2019s guidance<\/a> is quite straightforward about existing AI detectors. It says they have high error rates<\/a> and can lead instructors to falsely accuse students of misconduct.<\/p>\n

We\u2019ve already seen what that looks like in practice. Students have found themselves defending work they say they wrote themselves<\/a> after automated systems identified it as AI-generated. In one case documented by The Guardian<\/a>, a student\u2019s essay was flagged as entirely AI-generated despite the student saying they had only used approved spelling and grammar assistance. The appeal was eventually accepted.<\/p>\n

To be clear, Claude\u2019s watermark is fundamentally different. Conventional AI detectors<\/a> look at writing and essentially estimate whether an AI might have produced it. Anthropic is deliberately planting a detectable signal in Claude\u2019s output. In theory, that should make its system considerably more reliable. But reliability isn\u2019t the only problem here. Interpretation is.<\/p>\n

\"Using<\/p>\n

\t\tNadeem Sarwar \/ Digital Trends<\/p>\n

We\u2019re using AI to prove we didn\u2019t use AI<\/p>\n

Things have already reached a slightly ridiculous point.<\/p>\n

Students worried about AI detection are turning to so-called AI humanizers<\/a>, which rewrite text specifically to make it less likely to trigger detectors. Some students are even using these tools on work they wrote themselves because they\u2019re worried about false positives. Detector companies, naturally, are developing ways to identify humanizers.<\/p>\n

Read that again.<\/p>\n

A human can write something, worry that an AI will think an AI wrote it, feed it through another AI to make it look more human, and then have yet another system determine whether the AI made it look human.<\/p>\n

It\u2019s a technological ouroboros.<\/p>\n

Making Claude\u2019s watermark resilient enough to survive editing and translation is technically impressive. Previous research has shown that translation can defeat some text-watermarking techniques, so solving that weakness would represent meaningful progress.<\/p>\n

I just don\u2019t think making the signal harder to remove solves the more important problem.<\/p>\n

\"AI<\/p>\n

\t\tRachit Agarwal \/ Digital Trends<\/p>\n

A watermark needs context<\/p>\n

There are good reasons to watermark AI-generated content. It could help identify mass-produced misinformation, undisclosed synthetic text, or AI-written material that later ends up in training datasets.<\/p>\n

The problem is that AI assistants now do far more than generate content from scratch. People use them to translate text, proofread documents, summarize research, help with code, improve accessibility, or simply clean up an email before sending it. In that context, detecting AI involvement does not automatically tell you who actually created the work.<\/p>\n

All of those interactions involve AI to wildly different degrees. If Claude writes an essay from scratch, knowing that is useful. If Claude translates an essay someone spent three weeks researching and writing themselves, knowing Claude was involved tells you considerably less.<\/p>\n

The watermark may be perfectly capable of answering \u201cDid Claude touch this?\u201d My concern is what happens when people start treating the answer as proof of \u201cDid Claude write this?\u201d<\/p>\n

Anthropic can build the smartest watermark in the world. Unless the people using it understand that difference, I suspect we\u2019re going to have some problems.<\/p>\n","protected":false},"excerpt":{"rendered":"Anthropic wants to make AI-generated text easier to identify, and on paper, I have very little reason to…\n","protected":false},"author":2,"featured_media":125537,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[8],"tags":[53,3154,182,1221],"class_list":["post-141164","post","type-post","status-publish","format-standard","has-post-thumbnail","category-anthropic","tag-anthropic","tag-anthropic-claude","tag-claude","tag-computing"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/141164","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=141164"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/141164\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/125537"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=141164"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=141164"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=141164"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}