Executive Summary
Help Net Security reports prompt injection remains a major cause of agentic AI failures, citing CVE-2026-22708 in Cursor, where a poisoned execution environment let allowlisted commands deliver arbitrary payloads.
What Happened
Help Net Security reported that prompt injection remains a major cause of agentic AI security failures. The article cites CVE-2026-22708 affecting Cursor, where an attacker could poison the agent execution environment so allowlisted commands were used to deliver arbitrary payloads.
Affected Products
- Cursor
Vulnerability / Attack Technique
Poisoning the agent environment so allowlisted commands run payloads
Technical Mechanism
The attacker changes the environment the agent runs in, so commands on the allowlist execute attacker-controlled payloads.
AI Security Category
LLM01:2025 Prompt Injection
Part of the Prompt Injection topic.
Relevant Frameworks
RealGround Analysis
Allowlists only help if the environment they run in is trusted. RealGround analysis: protect environment variables, PATH and config that allowlisted commands depend on.
Recommended Controls
- Update Cursor
- Lock down environment variables and paths used by agent commands
Related RealGround Service
Secure AI Agent Build · or start with the free AI Agent Security Scanner.
Sources
Help Net Security: https://www.helpnetsecurity.com/2026/06/11/owasp-prompt-injection-ai-security-failures/
Summary and analysis written by RealGround from the source above; last updated 2026-10-18.