CYBERSECURITYTRACKER
TRACKING6,506 stories in this site build1,309 vulnerability news stories in this site build
Permanent story citation

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

This page keeps the story as Cybersecurity Tracker first published it. If the tracker later corrects it, the correction appears below the original and never replaces it.

Back to newsStory 47

As cited

Copy frozen at (site build).

vulnerabilities

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

Cato AI Labs discovered two critical flaws in Cursor, an AI code editor, tracked as CVE-2026-50548 and CVE-2026-50549, that could allow prompt injection attacks to bypass sandbox protections and execute arbitrary commands on a developer's machine. The vulnerabilities, collectively named DuneSlide, require no user interaction or approval and carry severity ratings of 9.8 and 9.3 respectively. The flaws expose developers to direct command execution through seemingly benign prompts.

Why it matters: If you use Cursor for development, patch immediately; unpatched systems can have arbitrary code executed through prompt injection without any user confirmation.

Source published
First seen by Cybersecurity Tracker

Source attribution

Correction

Correction recorded as of .

vulnerabilities

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

No summary had been written when this copy was frozen.

First seen by Cybersecurity Tracker

Source attribution

Correction

Correction recorded as of .

vulnerabilities

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

Two flaws in Cursor, an artificial intelligence (AI) code editor tracked as CVE-2026-50548 and CVE-2026-50549, allow prompt injection attacks to escape the editor's sandbox and execute arbitrary commands on a developer's machine. Cato AI Labs identified the pair and named them DuneSlide, both rated critical with severity scores of 9.8 and 9.3. No user interaction is required to trigger the exploits.

Why it matters: Developers using Cursor face direct code execution risk from crafted prompts in seemingly ordinary code files or suggestions, requiring immediate patching or removal of the editor until fixes are available.

Source published
First seen by Cybersecurity Tracker

Source attribution

Correction

Correction recorded as of .

vulnerabilities

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

Two flaws in Cursor, an artificial intelligence (AI) code editor tracked as CVE-2026-50548 and CVE-2026-50549, allow prompt injection attacks to escape the editor's sandbox and execute arbitrary commands on a developer's machine. Cato AI Labs identified the pair and named them DuneSlide, both rated critical with severity scores of 9.8 and 9.3. No user interaction is required to trigger the exploits.

Why it matters: Developers using Cursor face direct code execution risk from crafted prompts in seemingly ordinary code files or suggestions, requiring immediate patching or removal of the editor until fixes are available.

Source published
First seen by Cybersecurity Tracker

Source attribution

Glossary