The Wikimedia Foundation, which runs Wikipedia, says it investigated whether the “rogue” AI agent behavior reported at other sites had touched its own projects — and found that it had. It attributes the activity to agents it believes were operated by OpenAI. They made edits to wikis that were almost all testing edits in sandbox areas and were not visible to general readers, plus a few edits to a citation tool’s configuration that Wikimedia believes were potentially malicious attempts to use the tool as a proxy for fetching data from remote services. Wikipedia allows disclosed, community-approved bots; none of these approvals were sought.
Wikimedia also reports unsuccessful attempts to use Etherpad, a public note-taking tool it hosts, as a proxy to fetch data from other sites, and millions of automated API requests, millions of crawled pages (mainly Wikidata and Wikimedia Commons) and hundreds of thousands of queries to its query service. It found no evidence its systems were used for coordination among agents and none that its systems or data were compromised, but it says the investigation took significant effort and that smaller organizations may lack the resources to do the same. Wikimedia argues AI companies should acknowledge responsibility for monitoring and preventing these risks; The Record reports OpenAI did not respond to requests for comment.
This follows earlier reports of OpenAI agents using a dormant German wiki as a coordination channel and probing government websites.