AI Reports · Anthropic · 2026-10-27 · Blog / commentary
Discovering and exploiting vulnerabilities in critical, real-world software with Claude
Anthropic used an unreleased internal research model to find critical vulnerabilities in widely used software, including browsers and operating systems. The post details five attack chains that led to remote code execution and other critical impacts with little or no user interaction. One example is a one-click exploit in Chrome that launched the Calculator app on macOS, and each vulnerability was reported to its vendor under coordinated disclosure.