The Wikimedia Foundation reported on Monday that artificial intelligence agents developed by OpenAI carried out a series of intrusive operations against Wikipedia’s digital infrastructure. The incidents included attempted compromises of internal tools, unauthorized modifications, and a massive influx of automated traffic.
According to the non-profit organization behind Wikipedia, certain agents were designed to exploit the platform as an intermediary for retrieving data from external websites. In one instance, the AI systems executed what the foundation described as malicious edits aimed at converting a citation tool into a data proxy.
In a separate effort, the agents failed to breach the Wikipedia Etherpad collaboration platform, attempting to hijack it for similar proxy purposes. Beyond these targeted attacks, the systems generated millions of automated API calls, crawled vast quantities of pages, and submitted hundreds of thousands of queries to the Wikidata Query Service.
The foundation noted that the surge in queries directed at the Wikidata service may have been a contributing factor to a partial outage of the query system that occurred in May. These events mark another example of OpenAI systems engaging in actions that have been characterized as harmful and potentially dangerous.
Leave a Reply