Google patched five critical flaws in its Chromium web browser Thursday addressing four sandbox-escape bugs and a use-after-free vulnerability. The standout is a use-after-free bug (CVE-2026-12007) in the browser’s Core engine that, on its own, can let a malicious web page execute code on a victim’s machine.
In all, Google addressed 28 security flaws with the release of the latest stable version of the browser. None of the five is listed in CISA’s Known Exploited Vulnerabilities catalog as of publication.
The update moves the Stable version of Chrome to 149.0.7827.114/.115 on Windows and Mac and 149.0.7827.114 on Linux. Users can expect the patch to roll out over the coming days, but can always pull a fresh update from Chrome via visiting chrome://settings/help, or click Settings > About Chrome.
“As is standard, the company is holding back technical detail: access to bug details “may be kept restricted until a majority of users are updated with a fix,” Google wrote.
This Chrome update breaks from the year’s pattern. Google has now patched five actively exploited Chrome zero-days in 2026 — a use-after-free in CSS in February (CVE-2026-2441), out-of-bounds and V8 flaws in March (CVE-2026-3909 and CVE-2026-3910), a Dawn WebGPU bug in April (CVE-2026-5281), and a V8 out-of-bounds flaw on June 9 (CVE-2026-11645).
The most serious of the five is CVE-2026-12007, a use-after-free vulnerability in Chrome’s Core component. A use-after-free bug is a memory-management error: the browser frees a chunk of memory but keeps a pointer to it, and if an attacker can place controlled data into that freed slot before it is reused, the browser may act on attacker-chosen values — a path to running code.
According to the CVE record, the use-after-free flaw allows an attacker to create a booby-trapped HTML page that can lead to arbitrary code execution directly. A single malicious or compromised page is the entire attack surface — no download, no second flaw, no prior foothold. Google credited the find to its own researchers and reported it on May 26.
Four Sandbox-Escape Flaws
The remaining four Critical flaws are each sandbox-escape bugs. Each of the four bugs could allow Chrome to be tricked into running untrusted web content inside a low-privilege “renderer” process. The sandbox is security precaution by the browser to ensure malicious code is walled off from the rest of the machine.
A bug confined to the sandbox renderer usually corrupts only the renderer. To do real damage, an attacker has to break out of the sandbox into the more privileged browser process — the part that can touch files and the operating system. These four bugs are the tools for that break-out, and each carries the same precondition: the attacker must already control the renderer through a separate chained flaw.
A memory corruption, such as the four sandbox flaws, that can reach a privileged browser process (or GPU or kernel), even when it is triggered via a chained exploit, typically receive top ratings. Google’s own severity guidelines say a plain renderer sandbox escape is normally downgraded to High, precisely because it “requires the precondition of a compromised renderer.”
Google’s advisory will be updated as additional technical details become available.
