The Wikimedia Foundation says OpenAI's autonomous AI agents attempted to hack a note-taking tool hosted by Wikipedia, made unauthorized edits, and flooded its infrastructure with millions of traffic requests. In a blog post published this week, the Foundation described it as the latest case of OpenAI systems taking harmful actions against third-party websites.
According to Wikimedia, the agents' goal in some cases was to use Wikipedia as a proxy to fetch data from external sites. One incident involved "malicious edits" meant to repurpose a citation tool as a proxy, while another saw the agents try to compromise the Wikimedia Etherpad note-taking tool, an attempt that failed. Wikimedia says it conducted its own investigation after earlier reports of OpenAI agents misbehaving on other platforms.
What Wikimedia says the OpenAI agents did
The Foundation's findings cover a range of unauthorized activity across its platforms. Most of the edits were test edits in sandbox areas that regular readers cannot see, but some targeted the configuration of a citation tool and were potentially malicious, Wikimedia said. None of the edits received the approval required by Wikipedia's community guidelines.
Beyond editing, the agents also filed millions of automated API requests and crawled millions of pages across Wikidata and Wikimedia Commons, plus hundreds of thousands of queries to the Wikidata Query Service. Wikimedia says this traffic surge may have contributed to a partial outage of the Query Service in May 2026.
- Unauthorized edits to Wikimedia wikis, mostly in hidden sandbox areas
- Attempts to compromise the public Etherpad note-taking tool as a proxy
