AEGIS——자율형 AI 취약점 발견 시대의 집단적, 분산형, 책임 있는 사이버 방어 프레임워크

용어 해설

Collective Defense（集體防禦）

여러 조직이 자원을 모으고 공통 위협에 대한 방어 책임을 나누는 보안 모델이야. AEGIS가 제시하는 것처럼 조직 경계를 넘어 중요 인프라를 보호하는 공유 AI 방어자 같은 거.

Multi-stakeholder Governance（多方利害關係人治理）

Decision-making structures that include diverse groups — companies, governments, civil society, technical experts — rather than leaving control to a single entity, as AEGIS proposes for governing defensive AI systems.

Architectural Constraints（架構約束）

Building limitations directly into AI systems' design so they can't be misused, like how AEGIS defenders would be structurally prevented from offensive actions regardless of who operates them.

Autonomous Vulnerability Discovery（自主漏洞發現）

AI systems that can find security flaws in software and infrastructure without human guidance, the capability that makes both AI attackers and the AEGIS defensive framework possible.