Standard
OWASP LLM Top 10
OWASP Top 10 for Large Language Model Applications
Published by OWASP (Open Worldwide Application Security Project) Foundation, via the OWASP Gen AI Security Project · 2025 edition
A prioritized list of the most critical security risk categories specific to large language model (LLM) applications, raising awareness of LLM-specific threats such as prompt injection and excessive agency.
How it’s structured
- 10 numbered risk categories, each documented with a description, example attack scenarios, and mitigations
- Revised periodically as LLM application patterns evolve
Who it applies to
Developers and security teams building or assessing applications that integrate large language models.
How Pentforge uses it in an engagement
- Risk-category walkthrough during a penetration test or design review of an LLM-integrated application
- Findings referenced back to the specific OWASP LLM Top 10 category for a shared vocabulary with engineering teams
What you get
- A findings report mapped to the relevant OWASP LLM Top 10 categories
- Mitigation guidance scoped to the application's actual LLM integration points
Official source
Pentforge does not issue certificates
We assess against OWASP LLM Top 10 — that is a statement about the work, not a certification. Pentforge holds no certifications of its own; the credentials our practitioners hold individually are listed on About.
Request a scoping call