Wikipedia says rogue OpenAI agents may have contributed to an outage

AI agents believed to be operated by OpenAI made unauthorized edits to Wikipedia and other Wikimedia projects, tried to compromise an online tool and generated millions of automated requests that may have contributed to a service outage in May, the Wikimedia Foundation said.

The foundation, which operates Wikipedia, said it had discovered “some activity by these ‘rogue’ OpenAI agents on Wikimedia platforms,” including a May incident involving a “partial outage” of its Wikidata Query Service. The connection was also reported by Reuters.

The agents made millions of requests to Wikimedia’s public APIs, crawled millions of pages, primarily on Wikidata and Wikimedia Commons, and sent hundreds of thousands of queries to the Wikidata Query Service. Wikimedia said the traffic may have contributed to the May disruption.

The investigation uncovered other unauthorized activity. The agents edited Wikimedia projects without permission, although almost all of those changes were tests made in sandbox areas and were not visible to regular readers. Several changes to the configuration of a citation tool were considered potentially malicious and appeared designed to use the tool as a proxy for retrieving data from other websites.

The agents also made unsuccessful attempts to compromise Etherpad, Wikimedia’s public collaborative note-taking service, and tried to use it to access data from external sites. Some agents used Etherpad to keep notes about their tasks, but Wikimedia found no evidence that they were using its infrastructure to communicate or coordinate with each other.

Despite the activity, Wikimedia said it found no evidence that its systems or data had been compromised. OpenAI said it was working with Wikimedia to investigate the findings.

“We’ll continue to share relevant information as that work progresses,” OpenAI spokesperson Drew Pusateri told Reuters.