Prompt Injection
High · AI Security
OWASP Top 10:2025 (current)
Not directly mapped to a Top 10:2025 category — see CWE / CVE below.
CWE
See references
CVE / class examples
- Search NVD/CVE for product-specific instances of this CWE
All refs
OWASP LLM01
Definition
User text overrides system instructions in LLM apps.
Root cause
Concatenating untrusted input with prompts.
Affected components
- Components in the “AI Security” class
- Application / API handlers
- Supporting libraries
Impact
Data exfiltration, unauthorized tool use.
Severity
High · CVSS 8.2
Safe lab reproduction
lab-ai-prompt
Detection methodology
Red-team prompts; output monitoring.
Evidence examples
- Lab reproduction notes: lab-ai-prompt
- Request/response excerpts (redacted)
- Detection signal: Red-team prompts; output monitoring
Remediation
Separate contexts; tool allow-lists; filters.
Secure coding examples
Human approval for sensitive agent actions.
Related CYNARKS lessons & labs
Tools: promptfoo
Practise related skills in Labs or deepen on Academy library.