Google’s AI agent ‘Big Sleep’ foils cyberattack in groundbreaking first, says Pi

In a groundbreaking advancement for cybersecurity, Google’s CEO Sundar Pichai announced that the company’s AI agent, known as Big Sleep, successfully detected and intercepted a cyber exploit before it could be executed. This unprecedented achievement marks a significant milestone for artificial intelligence in the realm of threat prevention. Pichai shared the news via a post on X (formerly Twitter), stating, “New from our security teams: Our AI agent Big Sleep helped us detect and foil an imminent exploit. 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.” This event signifies a pivotal moment in cybersecurity, as AI transitions from merely identifying threats after a breach to actively preventing them from occurring. While Google has not provided specifics on when Big Sleep was launched or its operational timeline, Pichai’s announcement hints at an ongoing evolution of AI-powered defense mechanisms that will be integrated within Google’s ecosystem and available to cloud customers. With the frequency and severity of cyberattacks on the rise, the implementation of advanced AI like Big Sleep could soon become a standard practice for global IT security measures. Developed through a collaboration between Google DeepMind and Google Project Zero, Big Sleep is designed to autonomously discover unknown security vulnerabilities in software. In November 2024, Big Sleep made headlines by identifying its first real-world vulnerability, setting a new standard in AI-driven cybersecurity. Since then, it has uncovered several real-world flaws, exceeding expectations and pushing the boundaries of AI-assisted vulnerability detection. One of its most notable accomplishments was identifying a significant flaw in SQLite (CVE-2025-6965), which had previously only been known to cybercriminals. With assistance from Google Threat Intelligence, Big Sleep successfully anticipated a forthcoming exploitation attempt, preventing what Google claims is the first instance of AI proactively averting a vulnerability from being exploited in the wild. In addition to enhancing Google’s own security infrastructure, Big Sleep is now being utilized to fortify the security of commonly used open-source software, significantly contributing to the safety of the broader internet. These AI-driven agents represent a transformative shift in cybersecurity, allowing human teams to concentrate on more complex threats while vastly increasing their operational effectiveness and reach.