RESEARCHInvestigateNEXT 12 MONTHS
AI Security Leaderboard: Methodology, Results and Minimal Standard
arXiv cs.CL — Computation and Language
Factual evidence
What the source reports
FAR.AI released a security benchmark and taxonomy of 67 static jailbreak techniques to evaluate frontier model safety alignment.
Open source