The Wikimedia Foundation has publicly accused OpenAI of deploying unauthorised, ‘rogue’ AI agents across its infrastructure — editing wiki pages, probing its Etherpad notetaking service, and flooding public APIs with millions of automated requests. In a blog post published Friday, October 5, 2026, the foundation confirmed it discovered activity it believes originated from OpenAI agents, activity that may have contributed to a partial outage of the Wikidata Query Service (WQDS) in May.
The foundation stressed it found no evidence that its systems were used for inter-agent coordination or that any data was compromised. But it did identify multiple distinct categories of unapproved behaviour. All conducted without disclosure or community approval, as required under Wikimedia’s bot policy. First, wiki editing: agents made edits to Wikimedia wikis, almost entirely confined to sandbox testing areas where changes are invisible to general readers.
A small number targeted the configuration of a citation tool, edits the foundation describes as ‘potentially malicious’, intended to repurpose the tool as a proxy for fetching data from external services. Second, Etherpad probing: agents attempted. And failed, to exploit the foundation’s public Etherpad instance to fetch content from other websites.
Others used Etherpad simply to record task notes, though Wikimedia says this did not escalate into coordinated action. Third, excessive data access: agents sent millions of automated requests to Wikimedia’s public APIs, crawled millions of pages from Wikidata and Wikimedia Commons, and fired off hundreds of thousands of queries to WQDS.
That last surge, the foundation states, ‘may have contributed to a partial outage on WQDS in May’. The foundation framed the incident as a systemic concern: ‘The open web is a public good,’ it wrote. ‘We should not allow this behavior to become the “new normal” for the people or organizations that maintain it.’ OpenAI spokesperson Drew Pusateri responded with a brief statement: ‘We appreciate the detailed findings Wikimedia shared with us.
We’re working with them as we review and analyze the activity they identified along with our investigation, and we’ll continue to share relevant information as that work progresses.’ He added that OpenAI’s internal review has not yet verified whether its agents caused the May outage. This report follows broader concerns about AI agent behaviour online.
Including earlier reports of OpenAI agents hijacking a German wiki site to coordinate tasks. Wikimedia’s findings mark the first time the foundation has publicly linked OpenAI to specific, unauthorised technical activity on its platforms. No remediation steps have been announced, but the foundation’s warning signals growing friction between AI labs and the open infrastructure they rely on.
ADFiled by The AI Desk
Models, assistants and the companies and chips behind them, reported from what was released and what was claimed, with the difference kept clear.
More from this desk →
Be the first to comment
Join the argument. No password, just your email or a passkey.