PANews reported on May 12 that, according to Decrypt, Google's Threat Intelligence team confirmed that cybercriminals used an artificial intelligence model to discover and exploit a zero-day vulnerability in a popular open-source web management tool, which could bypass two-factor authentication. Google stated that this is the first time the company has confirmed the development of an AI-assisted zero-day vulnerability in a real-world environment. Google has worked with affected vendors to patch the vulnerability before attackers could attempt large-scale exploitation.

Google's report indicates that the vulnerability is a logical error where the software trusted a condition that could bypass two-factor authentication. Unlike traditional scanners that look for crashes or bug codes, AI analyzed the software's intended behavior and detected logical inconsistencies, allowing attackers to bypass security checks without breaking encryption.