Google has awarded a $250,000 bounty after security researchers using OpenAI’s GPT-5.6 Sol model crafted a full-chain exploit against Chrome. The vulnerability chain, developed by the XBOW team, linked multiple flaws to move from executing JavaScript in a Chrome tab to achieving remote code execution outside the renderer sandbox.
The exploit involved three distinct bugs, all discovered in early August 2026 and disclosed on October 13. It targeted Windows, though Google notes the flaws potentially impact nearly one billion Chrome users globally. All three vulnerabilities have since been patched.
The Three-Stage Bug Chain
First, the researchers uncovered a bug in V8, Chrome’s JavaScript engine. Within V8’s Maglev compiler, they found that separate JavaScript objects could share a mutable numeric value during a process called deoptimization. This allowed malicious code to corrupt memory inside V8’s protected areas.
The second vulnerability came from animated WebP image handling. The image decoder stored frame data from one image then reused it while processing modified data, leading to an out-of-bounds write. That error was leveraged to break out of the V8 sandbox.
The chain’s final bug was found in Crashpad, Chrome’s crash reporting system on Windows. It assumed memory reads during error recovery wouldn’t change, but the exploit introduced access modifications mid-retry. The result was an oversized memory read, letting attackers move code execution outside the renderer sandbox.
How The Attack Was Built, What It Means
The XBOW team leveraged GPT-5.6 Sol over roughly six days, costing about $40,000 in compute to build the full chain. Though AI played a role, human experts defined targets, reviewed outputs, and replaced earlier partial exploit chains when necessary to avoid known bugs.
Google’s reward program grants $250,000 for full-chain exploits arising from web content, provided they work on active Chrome releases and include a working demonstration under responsible disclosure. This case met all criteria.
Users are urged to update Chrome immediately and restart the browser once updates are applied. All three reported vulnerabilities have been patched, so staying current with versions is key.
On a broader scale, this event spotlights a growing trend: AI isn’t just a helper for documentation or prototyping—it’s becoming a capable tool for discovering deep security vulnerabilities. As AI agents gain strength in complex tasks, software teams are going to need tighter review processes, robust internal security testing, and improved mechanisms for AI-assisted bug discovery to reduce risk.
