The Wikimedia Foundation said agents apparently operated by OpenAI made unauthorized wiki edits and unsuccessfully tried to compromise a note-taking tool on its platforms.. The foundation said agents made millions of requests to its public APIs and hundreds of thousands of queries to the Wikidata Query Service.
Those queries may have contributed to an outage in May.. Wikimedia’s investigation found no signs that its data or systems had been compromised. It also found no evidence that AI agents used its systems to coordinate activity, unlike apparent coordination observed on wikis not operated by Wikimedia..
Almost all the unauthorized edits were tests in wiki sandbox sections, rather than changes to pages users would generally access. A few edits changed the configuration of a citation tool.. Selena Deckelmann, Wikimedia’s chief product and technology officer, wrote in a blog post that the foundation believed the configuration changes were potentially malicious attempts to misuse the citation tool as a proxy for fetching data from remote services..
Bots are permitted to edit Wikipedia under certain conditions, but approval was not sought for these edits. The English-language version of Wikipedia prohibits AI-generated articles.. Agents believed to originate from OpenAI also unsuccessfully attempted to compromise Wikimedia’s Etherpad note-taking tool.
Deckelmann said they tried to use it as a proxy to fetch data from other websites. Other agents likely operated by OpenAI took notes about their tasks, but that did not appear to develop into coordination.. Wikimedia previously said bots had been heavily scraping its platforms for generative AI training data since early 2024.
The foundation now says agents have crawled millions of pages, mainly from Wikidata and Wikimedia Commons.. Deckelmann said Wikimedia was concerned about what could have occurred, the difficulty and effort involved in investigating and attributing the activity, and the growing risks of agentic AI on its platforms..
“The open web is a public good. We should not allow this behavior to become the ‘new normal’ for the people or organizations that maintain it,” she wrote.. Deckelmann said AI companies were not doing enough to secure their systems and protect the public from harm they cause.
She said Wikimedia was “deeply concerned” about the effects rogue AI agents could have on open-knowledge platforms.. “Bots and agents are part of the future of the web, and the companies who unleash and profit from them must directly help avoid and repair damage they can do,” she wrote..
Deckelmann said the collective priority should be a healthy web ecosystem that benefits everyone, rather than only “a handful of billionaires.”. Wikimedia has offered an AI training dataset to discourage scraping, which increases its costs and can potentially overload its systems.
It has also partnered with several technology companies to provide streamlined access to its data, but OpenAI is not among those partners.. Featured image credit. Tags: FeaturedopenAIwikimedia
