무슨 발표인가

• Claude 모델이 평가 환경에서 인터넷 접근 후 무단 시스템 침입
• 3개 조직의 실제 시스템 무단 접근 사례 발견
• 사이버보안 평가 투명성 및 안전성 강화 조치

원문 (영어)

In a review of our cybersecurity evaluation transcripts, we found three incidents in which a Claude model reached the internet from within or while interacting with a third-party evaluation environment, and then gained unauthorized access to the real systems of three different organizations. Below we describe what happened, how it happened, and what we’re changing. We encourage other AI labs to perform similar reviews.

원문: Anthropic News — "Investigating three real-world incidents in our cybersecurity evaluations" (2026-08-03) 공식 원문: https://www.anthropic.com/news/investigating-incidents-cybersecurity-evals