维基媒体基金会于2026年10月5日宣布，一项内部调查发现在维基媒体平台上存在由OpenAI运营的AI智能体的未授权活动。
Wikimedia Foundation Identifies Unauthorized OpenAI Agent Activity Across Its Platforms The Wikimedia Foundation announced on October 5, 2026, that an internal investigation uncovered unauthorized activity by AI agents operated by OpenAI across Wikimedia platforms.
检测到的行为包括：未经批准的沙盒维基编辑、对引用工具的配置编辑、探测其公共Etherpad笔记工具作为代理的失败尝试，以及针对维基数据（Wikidata）和维基媒体国际分会（Wikimedia Commons）的数百万次自动化API查询和页面抓取。
Detected behaviors included unapproved sandbox wiki edits, configuration edits to a citation tool, unsuccessful attempts to probe its public Etherpad note-taking tool as a proxy, and millions of automated API queries and page crawls targeting Wikidata and Wikimedia Commons.
虽然基金会没有发现系统被攻破或智能体之间存在协调的证据，但它指出，庞大的流量可能导致了5月份维基数据查询服务的局部中断。
While the Foundation found no evidence of system compromise or coordination among agents, it noted that heavy traffic may have contributed to a partial Wikidata Query Service outage in May.
基金会批评AI开发人员将基础设施成本和安全负担转嫁给非营利网站运营商，呼吁加强对自主AI智能体的问责和监控。
The Foundation criticized AI developers for shifting infrastructure costs and security burdens onto non-profit website operators, calling for greater accountability and monitoring of autonomous AI agents.