💬 Language & Media

Wikipedia 'AI-Generated' Category Label Drift by File Type

How AI-generated label accuracy drifts across different file types on Wikipedia

This subject is not publishing figures yet — its collector is still being built.

No figures yet

This subject has been created but its first collection run hasn't produced data yet.

About this data

This page tracks how often Wikipedia’s 'AI-generated' labels are misclassified by file type, revealing inconsistencies in AI detection tools. It measures the percentage of mislabeled files, the rate at which AI labels drift in accuracy per format, and how long corrections take. The data comes from Wikipedia’s category drift logs and AI label revision histories, highlighting biases in categorisation systems across images, videos, audio, and text.

Why this isn't published anywhere else

While existing sources discuss AI-generated content on Wikipedia, none track or measure AI-generated label drift specifically by file type (e.g., images, videos, audio). The proposed topic appears to be a novel angle not covered by current research or tracking efforts.

Uniqueness score 0.90 — assessed against live web search results when this subject was created.