A new detection tool launched on June 29 threatens to escalate the fanfiction community's war against AI-generated stories - and human writers are in the line of fire. The anonymous X account @heatedrivalryai released a skin for Archive of Our Own (AO3) that claims to identify text created by Anthropic's Claude chatbot by uncovering a hidden code signature. The method is already under fire for being unreliable, and any author whose work ends up in AO3 could be falsely accused.
A long-simmering battle gets a new weapon
Fanfiction readers and writers have long traded tip sheets for spotting machine-generated prose. They point to em dashes, overly ornate passages, or what some call purple prose as red flags. But these gut-level markers never offered proof. The @heatedrivalryai skin claims to change that by catching a technical fingerprint Anthropic's tool leaves behind on the platform.
How the Claude code artifact works
The skin looks for a snippet of code - "font-claude-response-body" - that gets wrapped around text when it is pasted directly from Claude into AO3's editor. "When a Claude-generated response is pasted directly into AO3 from Claude, the text is wrapped by a Claude-injected code 'font-claude-response-body,'" the @heatedrivalryai account said. "Its presence indicates the use of Claude definitively." If a visitor has the skin installed, any AO3 page containing that code triggers a full red background. The original reporter confirmed the behavior: pasting directly from Claude turned the screen red; pasting the same story through a plain text editor did not.
Why this matters for writers
The skin's binary logic ignores nuance. A writer who drafts in Claude for structure, then reworks the text, could still trigger the flag if they paste from the chatbot window. The same goes for someone who uses a collaborative editing tool that inherits hidden markup. The result is a blunt instrument that pulls legitimate authors into the same dragnet as people passing off fully machine-generated work as their own. For writers trying to understand where AI fits into their process, separating fact from fear is critical. Practical resources like AI for Writers provide one way to learn what generative tools actually do - and what they don't.
Your membership also unlocks:
