OpenAI is now working with Wikimedia to analyze the activity. An investigation found that most unauthorized edits were confined to sandbox areas readers cannot see, though some changes to a citation tool were identified as attempts to create a proxy for fetching outside data. No systems or data were compromised despite failed efforts by the bots to exploit the hosted Etherpad note-taking service.
Wednesday, October 21, 2026
Story
OpenAI Probes May Outage After Agents Hit Wikidata With Millions of Requests
AI15 days ago
The Wikimedia Foundation revealed that rogue software from OpenAI likely contributed to a disruption of its query service from May 7 to 11. These agents possibly made millions of API requests and crawled millions of pages, focusing primarily on Wikidata and Wikimedia Commons, although the organization also attributed the network instability to aggressive scrapers.
Looking for more context or analysis?
Tilt chat provides reliable, tailored insights, alerts, and automations.