# AI-Powered Chrome Exploit Chain Developed in Six Days Security researchers leveraged LLMs to bypass Chrome sandboxes and execute code on Windows. - **Source:** [www.theregister.com - Articles (theregister.co.uk)](https://brifing.si/source/e6d4ec13-4d0e-407a-b96d-39609d76c3b1) - **Published:** 2026-10-14 - **Topic:** [Software & Engineering](https://brifing.si/c/engineering) - **Original article:** - **On Brifing:** https://brifing.si/s/56b264ff-4392-4169-b78d-2395223f412c ## AI-Accelerated Vulnerability Discovery Security startup Xbow successfully chained three zero-day vulnerabilities in Google Chrome to achieve code execution outside the browser's sandbox. Using OpenAI's GPT-5.6-Sol model, the team completed the entire exploit chain in just six days. This process, which historically required months of manual effort, highlights a significant shift in offensive security capabilities. ## Human-AI Collaborative Workflow The research team, including Brendan Dolan-Gavitt and Willem Pinckaers, utilized AI agents to search source code and test hypotheses. While the AI developed proofs of concept and exploit stages, human researchers provided critical context, selected targets, and challenged agent outputs. This hybrid approach ensured that complex security boundaries were effectively breached. ## Bypassing Chrome Sandbox Layers Chrome employs a nested sandbox design, including the V8 engine sandbox and an outer operating-system process sandbox. The exploit chain successfully crossed both boundaries in three distinct stages. By targeting specific memory-safety assumptions, the researchers moved from a V8 compiler bug to full native code execution on the underlying Windows machine. ## Technical Components of the Attack The attack sequence involved a V8 compiler bug in Maglev, an animated-WebP decoder vulnerability, and a memory-safety issue in Crashpad. The Maglev bug allowed read and write access within the pointer-compression cage. Subsequently, the team exploited Blink's trust assumptions to escape the V8 sandbox before triggering the final Crashpad overflow. ## Google's Response and Bounty Google addressed all three vulnerabilities and awarded the Xbow team a $250,000 bounty, representing their maximum payout for a full-chain exploit. Users are urged to verify they are running version 152.0.7977.82 or later. The security fix only becomes active once the browser is relaunched, ensuring the patch is fully applied to the system. ## Economic Efficiency of Exploitation The research effort cost approximately $40,000 in token charges, yielding a substantial return on investment given the $250,000 payout. This financial efficiency underscores how AI tools lower the barrier to entry for sophisticated cyber attacks. The project demonstrates that high-impact exploits are becoming increasingly accessible through the strategic application of LLMs. ## Historical Comparison of Timelines An intelligence-agency veteran noted that identifying and exploiting such vulnerabilities previously required six months of intensive work. Xbow achieved the same result in six days, marking a massive reduction in the time-to-exploit metric. This acceleration forces a reevaluation of how quickly software vendors must respond to emerging threats in production environments. ## Implications for Future Defense The success of this AI-driven exploit chain serves as a stark warning for software developers and security teams. As the bar for creating complex exploits continues to drop, defensive strategies must evolve to match the speed of AI-assisted attacks. Organizations should prioritize memory-safety and robust sandbox architectures to mitigate these risks. ## Sources covering this story - https://www.theregister.com/security/2026/10/14/security-researchers-spent-6-days-doing-their-chromework-built-1b-user-exploit/5303134 — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._