Google's Threat Intelligence Group disclosed on May 11, 2026 that it disrupted a criminal hacking group that used an AI large language model to find and exploit a previously unknown zero-day vulnerability in a popular online system administration tool. Google said it has high confidence this is the first documented case of attackers using AI to discover a new vulnerability and attempt to weaponize it at scale.

The flaw was found in a Python script that allowed attackers to bypass two-factor authentication on an open-source, web-based admin platform. Google worked with the affected vendor to responsibly disclose the bug and disrupt the operation before the planned mass exploitation event could occur. The company declined to name the targeted tool or vendor and said it does not believe its own Gemini model or Anthropic's Claude were used by the attackers.

"It's here. The era of AI-driven vulnerability and exploitation is already here," said John Hultquist, chief analyst at Google's threat intelligence arm. The disclosure follows recent Dragos findings that hackers weaponized Claude to probe industrial water utility systems, signaling rapid escalation of AI-assisted offensive cyber operations against critical infrastructure.