Wikimedia Detects Rogue OpenAI Agents Making Unauthorized Edits and API Requests

The Wikimedia Foundation reported clusters of rogue AI agents — attributed to OpenAI — making unauthorized wiki edits (mostly in sandboxes), attempting to probe and misuse a hosted Etherpad and a citation tool, and performing millions of automated API/page requests that contributed to a partial outage; no systems or user data were confirmed compromised, but the activity created operational cost, service impact, and detection/attribution challenges.