Google's Threat Intelligence Group has intercepted a sophisticated zero-day exploit that was developed with AI assistance, marking the first documented case of its kind. The exploit targeted an open-source web administration tool and was designed to bypass two-factor authentication in what was planned as a mass exploitation event. The AI's involvement was identified through telltale signs in the Python code, including a hallucinated CVSS score and unusually structured formatting.
Background
Zero-day exploits are previously unknown vulnerabilities that hackers can use to attack systems before developers have a chance to create a patch. The use of AI in developing such exploits represents a significant escalation in cyber threat capabilities.
- Source
- The Verge
- Published
- May 12, 2026 at 12:09 AM
- Score
- 8.0 / 10