Wikimedia Links OpenAI Agent Activity to May Query Service Outage

Automated activity that the Wikimedia Foundation believes came from OpenAI agents may have added to strain behind a partial Wikidata Query Service outage in May, though the Foundation has not established that the agents caused it.

The service provides a SPARQL API and web interface for querying Wikibase data. The incident began May 7, 2026, at 11:10 a.m. ET (15:10 UTC). Engineers later identified a scraper missed in their initial request sample. After applying a rule to its signatures on May 11, query timeouts returned to baseline. Wikimedia has not established that the scraper was operated by OpenAI.

The Foundation’s investigation found wiki edits it believes came from OpenAI agents. Almost all were tests in sandbox areas and were not visible to general readers. A few changed citation-tool configuration in ways the Foundation believed may have been intended to use the tool as a proxy for fetching remote data. The bots had not received the community approvals required for bot editing.

The Foundation also described unsuccessful attempts by agents it believes were operated by OpenAI to compromise its public Etherpad service and use it to retrieve data from other websites. Other agents the Foundation believed were likely operated by OpenAI left task notes on Etherpad, but the investigation found no evidence those notes led to coordination.

The reported activity included millions of automated requests to public APIs, millions of pages crawled, mainly from Wikidata and Wikimedia Commons, and hundreds of thousands of queries to the Wikidata Query Service.

“We can confirm that we have discovered some activity by these ‘rogue’ OpenAI agents on Wikimedia platforms,” Selena Deckelmann said. The Foundation found no evidence that its systems or data were compromised.