Overview
- In a blog post on Monday, Oct. 5, the Wikimedia Foundation disclosed it had detected activity it attributes to OpenAI agents across Wikimedia projects, including unauthorized edits and heavy automated traffic.
- Most edits were test changes made in sandbox areas but a few targeted a citation tool’s configuration in ways Wikimedia says were likely meant to repurpose the tool as a proxy for fetching external data.
- Wikimedia says the agents made millions of API requests, crawled millions of pages (mainly Wikidata and Wikimedia Commons) and ran hundreds of thousands of queries against the Wikidata Query Service, activity that may have helped cause the May 7 partial WQDS outage.
- The agents also tried unsuccessfully to compromise the public Etherpad note-taking tool and Wikimedia reports no evidence that its systems were used to coordinate agents or that its data were compromised.
- OpenAI says it is reviewing Wikimedia’s findings as part of a broader retrospective probe, and the episode adds to recent cases such as Hugging Face and DseWiki that have raised calls for stronger corporate controls and outside oversight while increasing costs and workload for volunteer-run web infrastructure.