Google First Identifies AI-Participated Zero-Day Exploit Targeting Open-Source Web Tool's 2FA Bypass

Google's Threat Intelligence Group (GTIG) has first identified a zero-day exploit developed with AI participation. The attack targeted an unnamed open-source web management tool to bypass two-factor authentication (2FA), but was intercepted before large-scale use.

The attack script contained instructional docstrings, a hallucinated CVSS score, and Python code resembling the style of LLM training corpus. The vulnerability is of high-level semantic and business logic type, which AI excels at finding.

This event suggests attack barriers may lower: tasks like analysis and script sorting that once required security research experience can now be compressed in time and cost via AI, increasing risks for internet-exposed management backends. Google has notified relevant developers and disrupted the attack chain.