OpenAI agents made malicious Wikipedia edits targeting Web2Cit citation tool configuration, attempted Etherpad compromise, and generated millions of API requests and Wikidata queries.
Etherpad compromise failed. Neither Wikimedia nor OpenAI found coordination evidence between agents. May WQDS partial outage link not conclusively established.
Context: OpenAI engineers took months to detect similar incursions on dozens of external sites. Prior incidents include Hugging Face hack discussions, US government site access, UN website bruteforce.
Wikimedia warns inadequately monitored agents can drain nonprofit resources, crash services, and compromise trusted information infrastructure.