344
Productivity & Workflow355
Automation & Workflow224
Software Development250
Marketing & Growth192
AI Infrastructure & MLOps174
Writing & Content Creation203
Data & Analytics141
Design & Creative169
Photography & Imaging156
Customer Support131
Sales & Outreach125
Voice & Speech135
Education & Learning131
Operations & Admin87
Fans on Archive of Our Own are sharing DIY tools that claim to detect AI written fanfiction, but the methods can be unreliable and cause disputes.
In short: Fanfiction communities are increasingly trying to detect and call out stories written with generative AI, and the tools people share can be shaky.
Some fanfiction readers and writers have been trading tips on how to spot stories made with tools like Claude and ChatGPT. These tips range from writing style clues, like heavy use of certain punctuation, to vague ideas like “purple prose” (writing that feels overly fancy or dramatic).
On June 29, an anonymous account on X, @heatedrivalryai, shared a new method aimed at Archive of Our Own, often called AO3. AO3 is a popular website where people post fanfiction and other fan works. The account posted a “skin,” which is basically a custom theme that changes how a page looks.
This skin is designed to flag text that may have been pasted directly from Anthropic’s Claude into AO3. It looks for a specific bit of behind-the-scenes web code, described as a “font-claude-response-body” marker. Think of it like checking whether a document still has a sticky note attached from the tool it came from. If the marker is present, the skin turns the page background red.
The Verge reports there are test posts on AO3 that users can visit to see the skin trigger, and the reporter also tested it against examples.
Tools like this can encourage people to police each other’s writing, even when the evidence is unclear or easy to avoid. The big question is whether AO3 communities will adopt shared rules for AI use, or whether accusations and false alarms will keep spreading.
Source: The Verge AI