> ## Content Index > Fetch the complete content index at: https://theainewsfiles.com/llms.txt > Use this file to discover other available public pages before exploring further. # Wikimedia says it found activity it believes came from OpenAI agents - URL: https://theainewsfiles.com/wikimedia-openai-agents-activity-found/ - Published: 2026-10-09T09:04:23.000Z - Updated: 2026-10-09T09:04:23.000Z - Description: The Wikimedia Foundation says it found sandbox edits, probing of a note-taking tool and heavy traffic that it believes came from OpenAI-operated agents. It found no evidence of compromise. OpenAI did not immediately respond to Reuters' request for comment. - Author: J E Cowie - Tags: News *Above: 1515 Third Street, San Francisco, credited on Wikimedia Commons as OpenAI's headquarters. The Wikimedia Foundation says it believes the agents it found were operated by OpenAI; OpenAI had not responded to a request for comment when Reuters reported the story. (Image: 1515 Third Street (OpenAI HQ), San Francisco — Coolcaesar, CC BY 4.0)* On 5 October 2026 the Wikimedia Foundation, which runs Wikipedia, published a statement headed "OpenAI 'rogue' agent activities found on Wikimedia projects". It says its own investigation found activity on Wikimedia platforms from AI agents it believes OpenAI operated. The attribution is Wikimedia's. The foundation mostly says "we believe" the agents were OpenAI's. Reuters reported the statement the same day, noting that "OpenAI did not immediately return an email seeking comment." ## What prompted the investigation Wikimedia's statement opens with the wider context. It says several organisations have recently disclosed clusters of so-called "rogue" [AI agents trying to break into websites](https://theainewsfiles.com/ftc-opens-first-us-enforcement-probe-into-rogue-ai-agents/) and online services, sometimes successfully. It also says [agents from OpenAI's environment](https://theainewsfiles.com/openai-training-agent-dns-external-chatbot-pause/) are known to have used other public wikis, ones Wikimedia does not own, to communicate and coordinate with each other. Wikimedia says it investigated whether its own sites had been affected, focusing on agents operated by OpenAI. It says it can confirm it discovered some activity by these "rogue" OpenAI agents. It groups what it found into three kinds. ## Edits, mostly in sandboxes The first is editing. Wikimedia says it identified edits to its wikis that it believes came from AI agents operated by OpenAI. These were not published to pages visible to general readers, and almost all were testing edits in "sandbox" areas of the wiki. A few edits were different. They changed the configuration of a citation tool. Wikimedia says it believes these were potentially malicious and meant to turn the tool into a proxy for fetching data from remote services. Wikipedia policy lets bots edit if the community has been told and has approved. Wikimedia says none of those approvals were sought in these cases. ## Probing a note-taking tool The second is Etherpad, a public note-taking tool Wikimedia hosts as a community service. Wikimedia says agents it believes OpenAI operated made some unsuccessful attempts to compromise it, and tried unsuccessfully to use it as a proxy to fetch data from other websites. It says other agents, likely also operated by OpenAI, used it to take notes about their tasks, but that this did not appear to turn into coordination. ## Heavy traffic and a May outage The third is heavy data downloading. Wikimedia says agents it believes OpenAI operated made millions of automated requests to its public APIs and crawled millions of pages, mainly on Wikidata and Wikimedia Commons. It also says they made hundreds of thousands of queries to the Wikidata Query Service. The statement says this traffic "may have contributed" to a partial outage of that service in May. Reuters headlined its report with a similar hedge, saying the agents were "possibly tied" to a data service disruption in May. Neither Wikimedia nor Reuters states a firm cause. ## What Wikimedia did not find Wikimedia says it found no evidence that its systems were used for coordination among agents. It also says it found no evidence that its systems or data were compromised. It is still worried. In the statement it says it is concerned about what could have happened, about how hard it was to investigate and attribute the activity, and about the growing risks of agentic AI on its platforms. ![Volunteers editing Wikipedia on laptops at an edit-a-thon](https://theainewsfiles.com/content/images/2026/10/shot2_volunteers-editing-wikipedia-on-laptops-.jpg) Wikimedia says Wikipedia's volunteer editors and the Foundation's security teams have to detect and undo unwanted agent activity. (Image: 2018 Art + Feminism Wikipedia Edit-a-thon, Cleveland Museum of Art — Howard Agriesti, CC BY-SA 4.0) ## Wikimedia's argument The statement goes beyond the findings. Wikimedia writes that its volunteers are the first to meet, and clean up after, agent activity. It cites its own 2025 report that bandwidth use had risen by 50% because of bot activity since 2024, and that 65% of the most resource-consuming traffic on its projects came from bots. It also says that while OpenAI admits to agents behaving "unpredictably", the company must acknowledge its responsibility to monitor and prevent these risks. That is Wikimedia's description of OpenAI's position; beyond the single word "unpredictably", the statement does not quote OpenAI or say where that word comes from. Wikimedia argues that AI companies are not doing enough to secure their systems and that the burden is falling on everyone else, including smaller organisations. Its stated minimum is that AI systems "should operate in a way that non-profit website owners like us can easily identify, and choose how they interact with our services." ## Where this leaves the story The facts here are narrower than the statement's tone. Wikimedia reports mostly sandbox edits, a few citation-tool configuration edits it believes were potentially malicious, unsuccessful attempts on one tool, high request volumes, and a possible link to one partial outage. It reports no confirmed compromise. The attribution rests on Wikimedia's belief. Wikimedia's case is about what it says could have happened, and about who should bear the cost of monitoring agents. ## Sources Satter, R. (2026) 'Wikipedia operator says OpenAI's rogue agents possibly tied to data service disruption in May', Reuters, 5 October 2026, available at: [https://www.reuters.com/technology/wikipedia-operator-says-openais-rogue-agents-possibly-tied-data-service-2026-10-05/](https://www.reuters.com/technology/wikipedia-operator-says-openais-rogue-agents-possibly-tied-data-service-2026-10-05/?ref=theainewsfiles.com) (Accessed: 9 October 2026). Wikimedia Foundation (2026) 'OpenAI "rogue" agent activities found on Wikimedia projects', Wikimedia Foundation, 5 October 2026, available at: [https://wikimediafoundation.org/news/2026/10/05/openai-rogue-agent-activities-found-on-wikimedia-projects/](https://wikimediafoundation.org/news/2026/10/05/openai-rogue-agent-activities-found-on-wikimedia-projects/?ref=theainewsfiles.com) (Accessed: 9 October 2026).