Wikipedia says “rogue” AI agents from OpenAI made edits without approval and sent millions of requests
The Wikimedia Foundation found no stolen data. It says the agents may have helped cause an outage in May, and that AI companies must do more.
Wikipedia is run by a non-profit called the Wikimedia Foundation. This week it said it had found AI “agents” from OpenAI acting on its websites without permission. An AI agent is a program that can browse the web and do tasks on its own.
The foundation looked into this after other groups reported similar problems. It says the agents made edits to its wikis. Almost all of them were test edits in practice areas that normal readers never see. A few edits changed the settings of a tool that adds citations. The foundation believes those edits were meant to misuse the tool.
The agents also tried, and failed, to break into a public note-taking tool the foundation hosts. And they made millions of automated requests for data. The foundation says this may have contributed to a partial outage of one of its data services in May.
The foundation found no sign that its systems or data were compromised. But it is worried. Bots already use a lot of its resources. Last year it reported that bot activity had raised its bandwidth use by 50 percent.
Its message to AI companies is blunt. It says they “must also acknowledge their responsibility to monitor and prevent these risks.” It wants bots to be easy to identify, so that website owners can choose how to deal with them.
The post ends with an invitation. The foundation asks everyone building the future of the web to help protect the open resources that make it possible.
Open it and check the claim yourself. That is the point.