An anonymous X account posted a skin for AO3 on June 29 that flags text containing a Claude‑injected code.

The skin adds a 'font-claude-response-body' wrapper to content copied directly from Claude into AO3’s editor, causing the page background to turn red when the code is present. Testing by the account and an independent reporter showed the red background appearing when a story was pasted straight from the chatbot and disappearing when the same text was edited before posting.

The code is only preserved if the text is transferred unchanged, so the tool cannot detect AI‑generated material that is edited in other programs. The red screen may indicate full AI authorship or only a brief passage that was checked with the model, and it does not reveal the extent of AI use.

Anthropic has not responded to requests for verification. Existing AI detection methods such as C2PA Content Credentials and Google’s SynthID apply to images, video and audio and do not cover copy‑pasted text. No universally reliable system for distinguishing AI from human writing currently exists.

AO3 offers a voluntary “Created Using Generative AI” tag for authors who wish to disclose AI use. The skin demonstrates a technical approach, but it may lead to false positives, and the community continues to rely on self‑reported signals rather than definitive verification.