Google announced that its AI agent, Big Sleep, developed by Google DeepMind and Google Project Zero, successfully detected and prevented the exploitation of a critical SQLite vulnerability (CVE-2025-6965) before hackers could launch an attack. This marks the first known instance of an AI agent directly foiling a live cyber exploit. The vulnerability was at risk of being exploited on millions of devices. Big Sleep actively searches for unknown software vulnerabilities and has full access to code and systems, enabling it to act swiftly. Experts caution that while AI agents like Big Sleep enhance cybersecurity, they also pose risks if unauthorized entities gain access to them. The incident highlights the evolving role of AI in cybersecurity defense and the importance of securing AI agents to prevent misuse.