Chrome 149 and Chrome 150 together patched 1,072 security bugs — a total Google says exceeds the number fixed across the previous 23 Chrome milestones combined.
How Google says it scaled vulnerability discovery with LLMs
According to Google, the company now employs large language models across the entire vulnerability management lifecycle: discovering flaws, reproducing reports, determining severity, assigning bugs to developers, generating candidate patches, and creating tests. Google began using LLMs to improve security fuzzing in 2023, then worked with Project Zero on a system called Naptime to give AI models specialized vulnerability-research tools. Later collaborations with Google DeepMind and Project Zero produced Big Sleep, an AI-powered vulnerability-discovery agent that found flaws in Chrome’s V8 JavaScript engine and graphics components.
Naptime, Big Sleep and a Gemini-powered agent
Google says it built on those early efforts by creating a Gemini-powered agent harness in early 2026 to search the broader Chrome codebase while reducing false positives. That multi-system approach reportedly located a range of bugs — including one Chrome sandbox escape that had remained in the codebase for more than 13 years. If exploited, Google said, the flaw could have allowed a compromised renderer to escape the sandbox and trick the browser into reading local files.
Automating triage, patch generation and testing
Google reports that automation now handles many traditionally manual steps. The company has automated vulnerability triage to filter spam and duplicates, reproduce proof-of-concept exploits, assign severity ratings, and route reports to the appropriate developers. After confirmation, fixing agents generate multiple potential patches; another agent evaluates those proposed fixes and produces additional information for developers to review. Google estimates this automated process saves hundreds of hours of developer time each month.
The company also emphasizes that automated AI workflows supplement rather than replace existing security testing. Fuzzing, which Google says remains effective at discovering complex vulnerabilities, continues to be part of the stack alongside AI-assisted discovery and testing.
Program changes, reporting volume and early outcomes
Google says it has seen a sharp increase in reports submitted through the Chrome Vulnerability Reward Program; by March 2026 the company had received more security bug reports than it did during all of 2025. To prioritize work, Google modified the program to give precedence to reports that add to what its automated tooling is already finding and processing. In May, Google reported that its systems prevented more than 20 vulnerabilities from reaching production, including one issue classified as critical.
Google is also encouraging developers to add SECURITY.md files that describe trust boundaries and threat models, a change intended to help AI systems better identify operations with security implications.
Patch delivery: two-week major releases, weekly updates and "dynamic patching"
Google acknowledges a practical tension: faster discovery shortens the time from discovery to public commit, but once a security fix is committed to Chrome’s public source code attackers can inspect the change and attempt to reverse-engineer the vulnerability before users receive the update. To shorten that window, Google is transitioning Chrome to a two-week major release cycle with weekly security updates and is piloting two security releases per week.
To reduce user disruption, Google is developing what it calls "dynamic patching" to allow Chrome to apply updates without restarting the browser. Starting with Chrome 150 on macOS, Google says the browser can automatically restart to apply a pending update when it is running in the background without any open windows. The stated long-term goal is to keep Chrome continuously updated through dynamic patching, automatic restarts during periods of inactivity, and improved session restoration.
What this means for technologists, procurement leaders, and end users
- Technologists and security teams: Expect faster, AI-assisted discovery and a heavier flow of candidate reports and patches. Google says automation will triage spam, reproduce exploits, assign severity, and route work, saving "hundreds of hours" of developer time each month.
- Affected enterprises and procurement leaders: The move to a two-week major release cycle, weekly security updates, and pilot twice-weekly security releases could require adjustments in update policies and testing cycles to avoid a backlog between public commits and deployed fixes.
- End users and the general public: Google’s aim to use dynamic patching and automatic background restarts starting with Chrome 150 on macOS is designed to reduce disruption and accelerate the arrival of fixes without requiring manual action.
Google’s account presents a single, concrete shift: AI is scaling the number of vulnerabilities Chrome teams can find and fix, and the company is changing both internal workflows and delivery mechanisms to keep pace. The trade-off is stark — faster discovery increases the urgency of getting fixes into users’ hands before public commits can be reverse-engineered — and Google’s near-term responses are a compressed release cadence and experiments in dynamic patching. Whether those steps will fully close the window attackers exploit after a public commit remains the immediate operational challenge Google has set for itself.
