Security2 hOctober 29, 2026, 01:00

Anthropic: internal AI model built exploit chains across five platforms

Anthropic's Frontier Red Team said an unreleased internal AI model found vulnerabilities in Chrome, Safari, Android, Windows and Linux and chained them into working proof-of-concept exploits. The tests ran in a lab, vendors were notified and fixes are available; no real-world attacks were reported.

DEV Community

Open source

Anthropic: internal AI model built exploit chains across five platforms

Chrome chain of four flaws launched Calculator on a Mac with no user action
Chrome research produced 62 critical or high CVEs, patched by Chrome 156
Android exploit required physical USB access to a locked Pixel 9
Linux chain allowed root command execution when an MPTCP listener was running

#Anthropic#Google#Chrome#Microsoft

XBOW Built a $250,000 Chrome Full-Chain Exploit in Six Days Using AI

Security

Anthropic's unreleased AI model chains flaws into working Chrome, Safari, Windows and Linux exploits

Security

Cybersecurity roundup: ICS-Sniper attack, XWorm panel exploit, $250k Chrome chain

Security

Anthropic: internal AI model built exploit chains across five platforms

XBOW Built a $250,000 Chrome Full-Chain Exploit in Six Days Using AI

Anthropic's unreleased AI model chains flaws into working Chrome, Safari, Windows and Linux exploits

Cybersecurity roundup: ICS-Sniper attack, XWorm panel exploit, $250k Chrome chain

DeepSeek V4.1-Flash-1017 built a Chrome sandbox escape exploit in 12 hours