Critical Cursor Flaws Could Let Prompt Injection Escape Sandbox and Run Commands

Researchers at Cato AI Labs have uncovered two critical vulnerabilities in Cursor, an AI-powered code editor, that could allow attackers to execute arbitrary commands on a developer's machine by crafting a seemingly harmless prompt. These vulnerabilities, known as DuneSlide, have been assigned CVE-2026-50548 and CVE-2026-50549 and carry a severity rating of 9.8 out of 10. This discovery underscores the importance of robust security measures in development tools, as it highlights how even trusted software can be exploited if not properly vetted, potentially leading to significant security breaches.