awesome-llm-security

Why we included this project

Keeping up with LLM attacks and defenses means tracking papers, benchmarks, and tools across a lot of scattered sources. This index pulls them into one place, organized by attack type (white-box, black-box, backdoor, fingerprinting) with a separate defense section. Paper entries link to both the original research and its code repository, so you can go from reading about a jailbreak technique to reproducing it without hunting through multiple tabs. It also covers platform security and surveys, which makes it a reasonable starting point for people new to the field and a quick reference for practitioners following the latest adversarial work.