The convergence of generative artificial intelligence and biotechnology has transformed the nature of bioterrorism threats and the methods used to address them. Using a descriptive-analytical approach and drawing on library sources, international instruments, and criminal law provisions, this study examines the dual-use role of generative AI in biosecurity. The findings indicate that large language models, biological design tools, and agentic AI systems can support genomic surveillance, environmental monitoring, biological sequence screening, genetic data protection, and non-proliferation mechanisms. At the same time, they may facilitate access to specialized knowledge, vulnerability identification, computational design, and cyber-bio attacks. These systems do not replace expert knowledge or laboratory infrastructure, but they can reduce informational and technical barriers to criminal conduct. Addressing such risks requires intelligent surveillance, protection of biological infrastructure and data, human oversight, and international cooperation. From a criminal law perspective, liability should be allocated according to each actor’s role, degree of control, level of awareness, foreseeability of risk, and breach of duties of care. Under Iranian law, large-scale biological attacks may be more clearly classified as efsad-e fel-arz than moharebeh, provided that all statutory and mental elements are established.