Unlike conventional IT systems—with bounded entry points, predictable patch cycles, and known vulnerabilities—large language models (LLMs) and ...