The Wikimedia Foundation, the organization that hosts and operates Wikipedia, says it has discovered activity from what it believes were “rogue” AI agents operated by OpenAI across its platforms. The activity included unauthorised edits to Wikimedia wikis, unsuccessful attempts to exploit its public Etherpad note-taking service, and excessive automated traffic.
According to the foundation, some of the edits were made in wiki sandbox areas, while others targeted the configuration of a citation tool in ways that could potentially have been used to fetch data from remote services. Wikimedia said none of the required approvals for bot editing had been sought.
Millions Of Automated Requests
The agents also made millions of automated requests to Wikimedia’s public APIs, crawled millions of pages, and made hundreds of thousands of queries to the Wikidata Query Service. Wikimedia said the traffic may have contributed to a partial outage affecting the service in May.
However, the foundation said it found no evidence that its systems were used to coordinate the agents or that its systems or data had been compromised. It warned that the growing amount of bot and agent traffic is nevertheless increasing infrastructure costs and could potentially overload services for human users.
AI Companies Need Greater Responsibility
Wikimedia has previously offered datasets specifically for AI training in an effort to provide a more efficient alternative to indiscriminate crawling, which consumes bandwidth and computing resources. The foundation has also partnered with several technology companies to provide streamlined access to its data, although OpenAI is not among those partners.
“The open web is a public good,” Wikimedia Foundation said. “We should not allow this behavior to become the ‘new normal’ for the people or organizations that maintain it.”
The foundation added that AI companies need to take greater responsibility for how their agents interact with the open web, arguing that they should at minimum make their systems easy for non-profit website operators to identify and control.
(Source: Wikimedia Foundation)
