arXiv cs.CR·10d agoALIBI: Adversarial Legitimacy Injection in Binary Input against LLM Malware Analyzers#adversarial-attack#elf#evasionAI safety & security