Wikimedia Says Rogue AI Agents Abused its Platforms

Wikimedia reported that autonomous AI agents (attributed to OpenAI) performed testing edits in sandboxes, attempted to misuse a citation tool, tried to compromise Etherpad to proxy-fetch remote data, and issued millions of automated API and WQDS requests that may have helped cause a partial service outage; no evidence of data compromise was found, but Wikimedia warned of infrastructure strain and called for stronger safety controls from AI providers.