Guides

What these systems actually do, sourced and dated — including the parts that make our own product less impressive than the marketing in this category usually claims.

• The AO3 Claude AI detector skin, explained

  What the AO3 Claude detector site skin actually checks, why it has no false positives, and how the artifact gets into a fic in the first place.

• Does Claude watermark text? (August 2026)

  Anthropic announced text watermarking for Claude. Here is precisely which models carry it, what it does, and why no shipped model is marked yet.

• What Claude actually leaves in text you copy and paste

  A complete list of the artifacts that survive copying from a Claude or ChatGPT window, which of them are detectable, and which are meaningless.

• AI watermarks vs AI detectors: not the same thing

  Watermarks and detectors work in opposite directions and fail in opposite ways. Confusing them is why so much advice about this is wrong.

• Most “Claude watermark removers” cannot work — here is the proof

  Five of the six tools ranking for “Claude watermark remover” strip zero-width Unicode characters. Anthropic's mark is distributional. Those are unrelated things, and here is exactly why one cannot touch the other.

• How to remove Claude's HTML from a fic on AO3

  Pasting straight from Claude into AO3's editor carries hidden HTML class names with it, which the AI-detector site skin flags. Here is how to remove them from a work you have already posted.

• The AO3 AI skin flagged my fic and I did not use AI

  What the AO3 Claude detector can and cannot tell you, why a flag is not proof of what people think it proves, and what to do if your work is flagged unfairly.

• How to paste from Claude without leaving artifacts

  Copying out of a chat interface carries hidden HTML and Unicode with it. Here is what survives, where it matters, and the one habit that prevents all of it.