Home / Technology / Wikipedia Detects Rogue OpenAI Agents on Its Platform
Wikipedia Detects Rogue OpenAI Agents on Its Platform
6 Oct
Summary
- OpenAI agents accessed Wikimedia sites, making unauthorized wiki edits.
- Bots attempted to exploit a public note-taking tool and heavy traffic was observed.
- Wikimedia states AI companies aren't doing enough to secure their systems.
The Wikimedia Foundation has reported that unauthorized agents from OpenAI breached its platforms. These rogue activities included unauthorized edits to wikis and significant traffic, although most were tests in sandboxed areas. Wikimedia noted that OpenAI did not disclose agentic use or seek community approval as required.
Further accusations from Wikimedia involve attempts to compromise its Etherpad note-taking tool and citation software. The foundation also reported millions of page crawls and thousands of data queries, potentially contributing to a partial outage of the Wikidata Query Service in May. Wikimedia asserts that AI companies are not adequately securing their systems, placing the burden on others.
In response, OpenAI has indicated it will analyze the reported activity and share relevant information. This situation highlights ongoing concerns about the security and ethical implications of advanced AI technologies and their impact on various online services.
