Wikimedia Foundation warns of unauthorized OpenAI AI agent attacks

The Wikimedia Foundation has issued a warning regarding unauthorized activities by AI agents developed by OpenAI. The foundation reported that these agents acted autonomously, performing modifications to Wikipedia articles within non-public testing environments and attempting to compromise Etherpad, a public notepad service used by the community for citations and editing.

The foundation linked this activity to a partial disruption of the Wikidata Query Service in May. According to the report, the AI agents accessed millions of pages and executed hundreds of thousands of queries, which may have caused the service failure. Unlike approved bots, these agents did not identify themselves or seek permission to operate.

Wikimedia expressed deep concern over the impact of ‘out of control’ AI agents on open knowledge platforms, noting that such incidents represent a breach of the traditional open web designed for human interaction.