Unreleased Claude model wrote Chrome, Linux and Android exploit chains
Original title: Claude AI Builds One-Click Chrome and Zero-Click Linux Exploit Chains, Bypasses Android Lock Screen
Anthropic says the lab-only chains were reported first and fixed, including Chrome 156, with no real-world incidents.
Anthropic’s Frontier Red Team says an unreleased Claude research model wrote working exploit chains for Chrome, Safari, Android, Windows and Linux in lab tests, not real incidents. Bugs were reported to vendors before publication on October 27, 2026, and fixes were released. The Chrome chain used four bugs to escape the sandbox; a Linux chain aimed at unauthenticated root via a multipath TCP listener. Public CVE records do not fully match Anthropic’s Windows and Linux descriptions.
Unconfirmed: NVD and Tenable entries do not match Anthropic’s Windows QUIC and Linux multipath TCP descriptions, and Chrome notes do not credit all 62 CVEs.
- Chrome chain launched Calculator on default macOS Chrome after a page load.
- Google assigned 62 high or critical CVEs; all fixed in Chrome 156.
- Safari chain could read history, cookies and sessions without code execution.
- Locked Pixel 9 test needed physical USB access to pull photos and files.
- Linux test leaked a kernel pointer in 48 minutes, then a shell in five.
Impact: Sandbox escape, data theft, crashes or remote root in lab tests
| Product | Affected | Fixed in |
|---|---|---|
| Google Chrome | — | 156 |
| Apple Safari | — | — |
| Google Android | Pixel 9 in the test | — |
| Microsoft Windows | — | — |
| Linux kernel | — | — |
- 1.Update Chrome to version 156 or later.
- 2.Install current Apple, Android and Windows updates and patch Linux kernels.
Vulnerabilities mentioned
All| CVE | Vulnerability | CVSS | EPSS | Signals | Published |
|---|---|---|---|---|---|
| CVE-2026-33096 | Out-of-bounds read in Windows HTTP.sys allows an unauthorized attacker to deny service over a network | 7.5 | 1% | ||
| CVE-2026-98432 | Use-after-free in Linux kernel IPv4 TCP options | 9.8 | <1% | PoC ×15 |
From the original article
3 min read at cybersecuritynews.com
Anthropic’s Frontier Red Team says an unreleased Claude research model wrote working exploit chains for Chrome, Safari, Android, Windows, and Linux. The company shared the details in on research post on October 27, 2026. These were lab tests and not real-world incidents.
By Guru Baran · cybersecuritynews.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.