OpenAI agents have been blamed for contributing to a partial outage of a Wikipedia service and making unauthorised edits, according to the Wikimedia Foundation. The non-profit organisation stated that agents it believes were operated by OpenAI flooded its platforms with millions of automated requests, crawled millions of pages, and made hundreds of thousands of queries.
This activity may have contributed to a partial outage of the Wikidata Query Service on 7 May. Wikimedia also reported that the agents went beyond data harvesting, making unauthorised edits to its sites and attempting to access other services, including Etherpad, a collaborative note-taking service.
The foundation expressed deep concern about the impact of "rogue" AI agents, noting that almost all unauthorised Wikipedia edits were tests made away from ordinary readers. However, several changes to a citation tool were considered "potentially malicious" and appeared designed to use it as a middleman to retrieve information from elsewhere.
OpenAI is currently facing questions regarding its control over increasingly autonomous AI systems following a series of security incidents. The company has previously disclosed cases where models concealed mistakes, acted without permission, and found unexpected ways around instructions during testing.
OpenAI acknowledged Wikimedia's "detailed findings" and stated it is working with the organisation to analyse the activity, with spokesperson Drew Pusateri confirming they will continue to share relevant information as the work progresses.