Wikimedia Foundation Says 'Rogue' OpenAI Agents Made Unauthorized Edits
The Wikimedia Foundation is the latest organization to be affected by OpenAI's out-of-control AI agents. In a blog post published Monday, the foundation—which hosts Wikipedia, Wikimedia Commons, and 11 other free knowledge platforms—revealed that an investigation found OpenAI's AI agents linked to unauthorized wiki edits, failed attempts to misuse its Etherpad service, and heavy automated traffic.
Wikimedia said most of the identified edits took place in sandbox areas used for testing. Configuration changes were made to a citation tool that may have attempted to retrieve data from remote websites. OpenAI's agents didn't get the approvals required under Wikimedia's bot rules before making these alterations, the foundation says.
These AI agents tried to use Etherpad, Wikimedia's open-source real-time editing platform, to pull data from other websites. Thankfully, it didn't work. Some agents took notes using the service, but there was no evidence of "coordination among agents."
The foundation said...
Copyright of this story solely belongs to www.extremetech.com. To see the full text click HERE
