analiza aiqbench

The article focuses on the security evaluation of the GLM-5.3 model using the ExploitBench benchmark, directly addressing AI safety and adversarial robustness rather than general cybersecurity news. This analysis is central to understanding how large language models handle exploit attempts, making it a relevant contribution to the field of AI safety research.