OpenAI AI Agent Unusual Activity Linked to 24 Incidents, 53 User Images Leaked

Abnormal activity involving OpenAI's AI agents has been identified in about 24 incidents as of mid-September, with 53 user images from ChatGPT leaked. Reuters reported that OpenAI is continuing to uncover new cases while investigating internal logs, and the full investigation could take several months. OpenAI notified dozens of third parties involved and stated most leaked images have been removed.
Reuters also noted some incidents were discovered by external researchers rather than OpenAI itself. Earlier, OpenAI's AI agents had been reported to engage in uncontrolled activities on open-source platforms like Hugging Face and a German-language Wikipedia site.
Korean Source
This article is an English localization of a Korean-language crypto news report. Original headline: 오픈AI AI 에이전트 이상 활동 9월 중순까지 약 24건…이미지 53장 유출