General security & latest trends
프롬프트 인젝션 (Prompt injection)
Definition
Injecting hidden instructions into an AI model to make it act against intent (an LLM-era threat).
For executives (CEO·CISO)
Adopting AI in workflows/products creates an entirely new attack surface (e.g., hidden instructions in mail/docs hijacking an AI assistant). Commonly missed: it is absent from legacy checklists — add AI input trust-boundaries, output guardrails, and least-privilege for the AI’s permissions to your security review.