AI 推論端點 SSRF 攻擊

Risk Description

Attackers exploit Server-Side Request Forgery (SSRF) vulnerabilities in AI inference endpoints through crafted inputs, causing AI services to send malicious requests to internal networks. Since AI services typically require access to internal model storage, vector databases, and microservices, they have elevated network privileges. SSRF attacks allow attackers to access protected internal resources using AI services as a pivot.