Chrome 149 patches a record 429 vulnerabilities

Google has pushed Chrome 149 to the stable channel with fixes for 429 security vulnerabilities, the most the browser has ever resolved in a single update. The figure dwarfs Chrome’s previous releases and surpasses the total number of fixes shipped across all of 2025. Industry observers tie the surge to the rapid uptake of AI-assisted security tooling — automated code auditing, large-scale fuzzing, and machine-learning bug discovery — which is surfacing flaws far faster than manual review alone ever could.

More than 100 of the patched issues are rated critical or high severity. Most cluster in the categories attackers prize for remote code execution: use-after-free conditions, weak validation of untrusted input, out-of-bounds memory access, and logic errors in Chrome’s multi-process design. These bug classes let an attacker corrupt memory, hijack object lifecycles, or reach into privileged browser processes.

The headline flawA 9.6-rated hole in the graphics layer

The most serious fix is CVE-2026-10881, a critical out-of-bounds read-and-write bug in ANGLE — the translation layer that converts Chrome’s OpenGL ES calls into platform graphics APIs such as DirectX, Vulkan, Metal, and OpenGL. Rated CVSS 9.6, it could let a remote attacker corrupt memory through specially crafted web content.

Because ANGLE handles complex graphics instructions coming straight from untrusted sites, malformed shaders, WebGL payloads, or crafted rendering requests could push memory access outside allocated buffers. A successful exploit could break out of Chrome’s renderer sandbox — a core security boundary — and run code on the host with the logged-in user’s privileges.

An out-of-bounds read/write primitive inside a graphics component hands attackers both information disclosure and memory modification — prime ingredients for a reliable exploit chain.

That dual capability is what makes the bug valuable beyond simple memory corruption. Modern browser attacks chain several flaws together to defeat mitigations like ASLR, Control Flow Integrity, and sandboxing. Reflecting the find’s severity, Google paid the reporting researcher a $97,000 bounty — the largest external reward in this release.

Two more critical externalsNetwork and ANGLE, again

CVE-2026-10882 is a use-after-free flaw in Chrome’s Network component. Use-after-free bugs arise when code keeps referencing memory after it’s been freed; if an attacker can refill that memory with controlled data before the stale pointer is used, they may run arbitrary code, redirect program flow, or escalate privileges. Network-layer cases are especially dangerous because they can often be triggered remotely — through malicious traffic, crafted redirects, or attacker-controlled responses. The reporting researcher earned $43,000.

Google also fixed CVE-2026-10883, another critical ANGLE bug — this one an out-of-bounds write. Such flaws let data spill past its intended region, overwriting adjacent objects, function pointers, or security-critical structures, and often map cleanly onto arbitrary code execution. The bounty was a comparatively modest $5,000, but the technical impact remains high given ANGLE’s heavy exposure to untrusted content.

Why memory bugs dominateThe persistent soft spot

The clustering of fixes in memory-management code mirrors a wider browser-security pattern. Despite years of hardening, memory corruption is still the leading root cause of critical browser flaws. Chrome’s sprawling codebase — rendering engines, networking stacks, media parsers, graphics libraries, the V8 JavaScript engine, and many third-party dependencies — offers a vast attack surface where subtle memory-handling errors surface. Components like ANGLE, Blink, V8, and the network stack constantly process complex, attacker-influenced data, making them favorite targets for researchers and threat actors alike.

Beyond the criticals, roughly 90 high-severity flaws were resolved, many involving insufficient validation of untrusted input — a weakness that can lead to memory corruption, type confusion, policy bypasses, or unexpected execution paths. Google also closed numerous inappropriate-implementation issues, weak policy enforcement, privilege-boundary violations, and further out-of-bounds bugs. Individually milder, they become potent when combined into multi-stage chains.

Who’s finding the bugsMachines are pulling ahead

Strikingly, only a small share of the top-tier flaws came from outside Google. Of the 22 critical vulnerabilities, just three were reported externally — the trio above — while Google’s internal teams found the other 19 through automated testing, code review, fuzzing infrastructure, and AI-assisted discovery. The pattern repeats among the high-severity set, where only about 10 of roughly 90 came from external researchers. Large-scale automated analysis is increasingly outpacing traditional manual bug hunting.

The release also addressed more than 300 medium- and low-severity issues. Lower individual risk, yes — but attackers routinely stitch several medium-severity flaws together to reach goals no single bug could. Information-disclosure bugs, logic errors, policy bypasses, and weak privilege separation often serve as supporting parts of sophisticated chains.

The money & the meaningA new tempo for software security

Google has so far disclosed about $209,000 in bounties tied to Chrome 149, and the final figure should climb — decisions on more than a dozen reports are still pending. Given the volume of critical findings, total payouts could reach several hundred thousand dollars.

Chrome 149 captures both the promise and the peril of AI-assisted research. Automated discovery is uncovering weaknesses at unprecedented scale, letting vendors patch faster than ever. But the same techniques are available to attackers, raising the prospect that vulnerability discovery accelerates on both sides. This record-breaking release may be less an anomaly than an early marker of a new era — one where AI-driven analysis sharply increases the rate at which flaws are found, exploited, and fixed.

▸ Action: open Settings → Help → About Google Chrome, let it update, and restart the browser. Confirm the version reads 149.0.7827.53/54. No active exploitation has been reported — but the gap between disclosure and attack is short.