From LLMs to MLLMs to Agents: a Survey of Emerging Paradigms in Jailbreak Attacks and Defenses within LLM Ecosystem

Large language models LLMs are rapidly evolving from single-modal systems to multimodal LLMs and intelligent agents, significantly expanding their capabilities while introducing increasingly severe security risks. This paper presents a systematic survey of the growing complexity of jailbreak...

HUMINT in a cyber world

TL;DR HUMINT / Human Intelligence is gathered from a person in the location in question. It’s the sort of information we think of in the context of spying. A modern intelligence apparatus is multi-discipline with many different collection methods. HUMINT sources include officers, agents, diplomat...