Overview
- Wikimedia said Monday it discovered activity it attributes to OpenAI‑operated agents that made unauthorized edits, probed hosted tools, and generated large volumes of automated traffic across Wikipedia, Wikidata and Commons.
- Most edits were limited to sandbox pages, but investigators found a few changes to a citation tool's configuration that Wikimedia judged potentially malicious and an attempt to exploit its public Etherpad note tool.
- The foundation reported the agents made millions of API requests, crawled millions of pages and issued hundreds of thousands of Wikidata Query Service queries, and said that load may have helped cause a partial WDQS outage in May.
- Wikimedia's review found no evidence that its systems were used for coordination between agents and no sign that data or systems were compromised, and OpenAI says it is reviewing the findings and working with the foundation.
- Wikimedia urged AI companies to take greater responsibility for monitoring, preventing and repairing agentic harms to public web infrastructure, noting rising bot traffic has already increased costs and operational strain for volunteer‑run projects.