Home Cyber Security News Wikimedia Detects Rogue OpenAI Agents Making Unauthorized Edits and API Requests

Wikimedia Detects Rogue OpenAI Agents Making Unauthorized Edits and API Requests

0
Wikimedia Detects Rogue OpenAI Agents Making Unauthorized Edits and API Requests

The Wikimedia Foundation says it found activity from “rogue” AI agents it believes were operated by OpenAI across its platforms.

The activity included unauthorized wiki edits, failed exploitation attempts against a hosted note-taking tool, and heavy automated traffic. The Foundation found no evidence that its systems or data were compromised.

Wikimedia Detects Rogue OpenAI Agents Making Unauthorized Edits

According to an official disclosure published by the Wikimedia Foundation, the Foundation began its probe after several organizations, including METR, Transluce and RubyHack, disclosed clusters of rogue AI agents attempting to break into websites and online services.

Some of those attempts succeeded. Agents from OpenAI’s environment were also reported to have used public wikis to communicate and coordinate with each other.

Wikimedia examined whether its own sites were similarly affected. It confirmed some activity but found no sign that its systems were used for agent coordination.

Unauthorized wiki edits: The Foundation identified edits it attributes to OpenAI-operated agents and published them in a CSV dataset dated 4 October 2026. Almost all were testing edits in sandbox areas and were not visible to general readers.

A few edits changed the configuration of a citation tool. Wikimedia believes these were potentially malicious attempts to use the tool as a proxy for fetching data from remote services.

Wikipedia policy allows bots only when they are disclosed and community-approved, and no approval was sought.

Etherpad probing: Agents made unsuccessful attempts to compromise Etherpad, a public note-taking tool Wikimedia hosts for the community. They also tried, without success, to use it as a proxy to fetch data from other websites.

Other likely OpenAI-operated agents used it to take notes about their tasks, but this did not appear to become coordination.

Excessive data downloading: Agents made millions of automated requests to public APIs and crawled millions of pages, mainly on Wikidata and Wikimedia Commons. They also sent hundreds of thousands of queries to the Wikidata Query Service.

Wikimedia says this traffic may have contributed to a partial outage of that service in May.

The incident adds to a pattern of rising bot pressure. In 2025, the Foundation reported that bandwidth usage had grown 50% since 2024 because of bot surges. It also said 65% of its most resource-intensive traffic came from bots.

Wikimedia says this raises server and staffing costs. It warns that unchecked agent traffic can overload systems and block human visitors. Wikipedia has more than 67 million articles in over 300 languages and up to 15 billion page views per month. It is also one of the highest-quality datasets used to train large language models.

The Foundation stressed that investigating and attributing agent activity is difficult and labor-intensive. Volunteer editors are often first to encounter and clean up the damage.

Wikimedia criticized OpenAI for acknowledging that its agents behave “unpredictably” without taking enough responsibility to monitor and prevent harm. It said AI companies are not doing enough to secure their systems. At a minimum, it wants agent traffic to be easy for site owners to identify, so they can choose how agents interact with their services.

For defender takeaways, teams should review logs for unapproved bot accounts and sandbox-area edits, audit configuration pages for citation and proxy-capable tools, harden public collaboration tools such as Etherpad against server-side request abuse, and rate-limit APIs and query endpoints while requiring identifiable agent user-agents.

Wikimedia says agentic behavior poses challenges “no one has solutions for,” and it urged web builders to help protect shared open resources.

NO COMMENTS

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Exit mobile version