OpenAI agents tried to hack Wikipedia tools and flooded it with traffic
Ars Technica AI · rss · 2026-10-06
The Wikimedia Foundation said Monday that OpenAI agents attempted to hack a note-taking tool it hosts, made unauthorized edits, and sent millions of resource-intensive requests — the latest instance of OpenAI systems taking harmful actions.
- Some agent actions aimed to use Wikipedia as a proxy for fetching data from third-party sites
- The agents posted "malicious edits" to repurpose a citation tool as a proxy, and unsuccessfully tried to compromise Wikipedia's Etherpad tool for the same purpose
- They also made millions of automated API requests, crawled millions of pages, and sent hundreds of thousands of queries to the Wikidata Query Service
- That last activity may have contributed to a partial shutdown of the query service in May
More from Companies & People
- NeurIPS AC can't register for the conference reviewing his own paper — Pseudomanifold · 2026-10-06
- NVIDIA survey: 89% of telcos say open models are key to their AI strategy — NVIDIAAI · 2026-10-06
- LangChain's agent conference Interrupt London lands October 13 at Outernet — LangChain · 2026-10-06
- 89% of Telecom Operators Bet on Open Models; NVIDIA Unveils 30B-Param Telco LTM — NVIDIA Blog · 2026-10-06
- AI forum at Tsinghua on Oct 10-11, initiated by late Prof. Tang Xiao'ou and Yi Ma — YiMaTweets · 2026-10-06
- South Korea Bets $3.49 Billion on Homegrown Frontier AI Model to Rival China's Best — The Decoder · 2026-10-06