OpenAI agent sparks chaos again; Wikimedia flags millions of requests, unauthorised edits

AI agents are being designed to browse websites, collect information and complete tasks with increasing independence. But their activity across the open web can also put additional pressure on the services they interact with. The Wikimedia Foundation says it has identified activity that it believes came from OpenAI-operated agents on its platforms. This included unauthorised edits to Wikimedia wikis, attempts to use its public note-taking service to access other websites and millions of automated requests to its APIs. 

Must Read: OpenAI plans to watermark ChatGPT and Codex text output in EU amid AI transparency push

The activity also involved millions of crawled pages, mainly across Wikidata and Wikimedia Commons, along with hundreds of thousands of queries to the Wikidata Query Service. Wikimedia says this traffic may have contributed to a partial outage of the service in May. 

What did the OpenAI agents do on Wikimedia?

According to Wikimedia, it identified edits that it believes were made by AI agents operated by OpenAI. Most of these were testing changes made in sandbox areas, meaning they were not published on pages visible to general readers. A few edits changed the configuration of a citation tool, which Wikimedia believes could have been intended to misuse the tool as a proxy for fetching data from other services. The foundation also noted that the required bot approvals were not obtained for these edits.

Must Read: Superintelligence could turn 'super dangerous' says SoftBank's Masayoshi Son

Wikimedia also found unsuccessful attempts involving Etherpad, its public note-taking service. Some agents tried to use the tool to fetch information from other websites, while other agents believed to be linked to OpenAI used it to record notes about their tasks. However, the foundation said it found no evidence that this activity turned into coordination between agents.

Millions of requests put additional pressure on services

The investigation found millions of automated requests to Wikimedia's public APIs and millions of pages being crawled. The agents also generated hundreds of thousands of queries on the Wikidata Query Service, with Wikimedia saying this traffic may have contributed to its partial outage in May.

The foundation says this comes amid a broader rise in automated traffic. Wikimedia reported that its bandwidth usage increased 50% in 2025 because of rising bot activity since 2024, while bots accounted for 65% of its most resource-intensive traffic. Wikipedia now has more than 67 million articles across more than 300 languages and receives up to 15 billion page views every month.