When a Fan Detector Turned AO3 Red: What Happened in the Heated Rivalry Fandom

An anonymous X account called @heatedrivalryai posted a document on June 29, 2026 that exposed hidden Claude-specific HTML artifacts embedded in some stories on Archive of Our Own (AO3). The document listed 38 works and included a simple browser skin that highlights affected pages by turning their background red, a visual cue that many readers found hard to ignore. What the skin reveals is not a proof of total automation but an artefact left when text is copied directly from the Anthropic model Claude into AO3’s editor; the artifact vanishes if the same text is routed through a word processor before posting.

Because the detection depends on that specific workflow, motivated actors can evade it, and innocent users may be flagged if they used Claude only for grammar or translation. Nevertheless, the skin accelerated a public reckoning: accusations circulated quickly, and some authors received abusive messages, deactivated accounts, or temporarily made works private. At the same time, a few creators acknowledged limited AI assistance and added tags or notes to clarify their process.

Archive of Our Own’s policy perspective remains preservationist; the site’s administrators maintain that fanworks, including AI-assisted pieces, fall within the archive’s remit while moderation focuses on abuse and plagiarism. Consequently, the debate has shifted from simple detection to questions of disclosure, trust, and community norms — issues that will shape how fandom adapts to writing tools that are now widely available. Scholars and platform operators will likely watch these debates closely as practices evolve. Change will not be immediate.