The North News
Chandigarh, July 16
Google CEO Sundar Pichai announced on July 15, “We believe this is a first for an AI agent definitely not the last giving cybersecurity defenders new tools to stop threats before they’re widespread.” In a major breakthrough for cybersecurity, Google’s AI agent Big Sleep successfully detected and stopped an imminent cyber exploit before it could be launched a first-of-its-kind feat for artificial intelligence in real-world threat prevention. Pichai called the achievement a milestone that signals a new era of proactive AI-driven defense.
Developed by Google DeepMind in collaboration with Project Zero, Big Sleep is designed to autonomously detect unknown vulnerabilities in software. Since identifying its first real-world flaw in November 2024, the AI agent has uncovered multiple critical issues, including a high-risk vulnerability in SQLite (CVE-2025-6965) that had only been known to malicious actors. With support from Google Threat Intelligence, Big Sleep predicted an imminent exploitation attempt and successfully foiled it, marking what Google believes to be the first time AI has actively prevented a cyberattack in the wild.
While details about Big Sleep’s deployment remain undisclosed, its impact is already being felt beyond Google’s infrastructure. The AI is now aiding in the protection of widely used open-source software, helping improve internet-wide security. As cyberattacks become more sophisticated and frequent, AI-powered agents like Big Sleep are poised to become standard tools in global cybersecurity allowing human teams to focus on complex threats while expanding detection and prevention capabilities.
