Google Detects First AI-Generated Zero-Day Exploit Designed to Bypass Two-Factor Authentication
Tags AI · Security · Infrastructure
Google's Threat Intelligence Group identified and disrupted the first known zero-day exploit developed with AI assistance — a Python script targeting a 2FA bypass vulnerability in an unnamed open-source web administration tool. Evidence of AI involvement includes a hallucinated CVSS score, educational docstrings, and textbook Pythonic formatting characteristic of LLM training data. The criminal group was planning mass exploitation. GTIG also observed Chinese state-sponsored actors using persona-driven jailbreaking to find vulnerabilities in embedded devices.
Technical significance
This marks the confirmed start of AI-assisted offensive cybersecurity operations at scale. The exploit targeted a semantic logic flaw — exactly the type of vulnerability LLMs excel at finding. For defenders, this means vulnerability scanning and patching cycles must accelerate. The use of persona-driven jailbreaking to find vulnerabilities is a technique that existing AI safety measures are not designed to prevent.