AI Interactive Tools
OWASP LLM Top 10 Self-Assessment
40+ control questions across 10 risks; weighted score + remediation for top 12 gaps.
- OWASP LLM Top 10
- OWASP Foundation's Top 10 risk list for Large Language Model Applications (2025 edition); defines 10 main security risks from prompt injection to unbounded consumption with remediation guidance.
- Also known as: LLM security, AI red team, OWASP LLM01-10
LLM01Prompt InjectionImpact: 5/5ATLAS: 3User or external content overriding the model's system instructions through crafted input.
LLM01-Q1.Do you separate system and user prompts with clear delimiters?
LLM01-Q2.Do you mark content from external sources (web, PDF, email) as 'untrusted' in the system message?
LLM01-Q3.Are sensitive actions gated by out-of-band user confirmation?
LLM01-Q4.Do you run a prompt injection test suite continuously (in CI)?
LLM01-Q5.Is agent tool access granted on least privilege?
LLM02Sensitive Information DisclosureImpact: 5/5ATLAS: 3Model inadvertently revealing system prompts, customer data, secrets or internal context.
LLM02-Q1.Are you certain no secrets/API keys/credentials reside in system prompts?
LLM02-Q2.Do you apply PII redaction on model request/response logs?
LLM02-Q3.Is there monitoring for PII/secret leakage in output?
LLM02-Q4.Is there a guardrail blocking attempts to reveal the system prompt?
LLM03Supply Chain VulnerabilitiesImpact: 4/5ATLAS: 3Vulnerabilities from third-party models, datasets, plugins or libraries.
LLM03-Q1.Are third-party models' provenance, licence and training data transparency verified?
LLM03-Q2.Is an SBOM maintained for the ML pipeline?
LLM03-Q3.Do plugin/tool additions go through security review?
LLM03-Q4.Is vendor incident notification time contractually max 48 hours?
LLM04Data and Model PoisoningImpact: 5/5ATLAS: 3Malicious content injected into training, fine-tuning or RAG data.
LLM04-Q1.Do documents loaded into the RAG vector DB go through an approval workflow?
LLM04-Q2.Is licence + copyright verification done on fine-tune data sources?
LLM04-Q3.Is anomaly detection (e.g. cosine outliers) running on the vector DB?
LLM04-Q4.Is rollback procedure documented for model/RAG?
LLM05Improper Output HandlingImpact: 4/5ATLAS: 2Rendering model output exposes XSS, SQLi, command injection.
LLM05-Q1.Is XSS sanitisation applied when rendering LLM output?
LLM05-Q2.Are LLM-sourced code/SQL/Shell outputs never directly executed?
LLM05-Q3.Is model JSON output validated against schema?
LLM06Excessive AgencyImpact: 5/5ATLAS: 2Agent given more authority than needed; irreversible/high-impact action rights.
LLM06-Q1.Is each agent's scope documented (allowlist)?
LLM06-Q2.Do irreversible actions require HITL approval?
LLM06-Q3.Are agent actions rate-limited?
LLM06-Q4.Is there an emergency kill switch?
LLM07System Prompt LeakageImpact: 3/5ATLAS: 2System prompt content leaked via reverse engineering.
LLM07-Q1.Does the system prompt avoid business logic / secrets?
LLM07-Q2.Are prompts versioned + auditable?
LLM08Vector and Embedding WeaknessesImpact: 4/5ATLAS: 2Vector DB security, embedding inversion, cross-tenant data leakage.
LLM08-Q1.Is there a separate vector namespace per tenant in multi-tenant use?
LLM08-Q2.Are embedding endpoints protected with authentication?
LLM08-Q3.Is vector data encrypted at-rest + in-transit?
LLM09Misinformation (Hallucination)Impact: 4/5ATLAS: 2Model confidently producing false information; user/customer risk.
LLM09-Q1.Is source citation mandatory when generative AI output is used with RAG?
LLM09-Q2.Is fact-checking process documented for public-facing content?
LLM09-Q3.Are model's low-confidence answers flagged to the user?
LLM10Unbounded Consumption (DoS / Cost)Impact: 3/5ATLAS: 2Unbounded token consumption; attack or bug causing excessive cost/infra usage.
LLM10-Q1.Is there an hourly per-user token limit?
LLM10-Q2.Are monthly cost alarms + kill-switch configured?
LLM10-Q3.Is there truncation/refusal logic for excessively long input?
Score
OWASP LLM Top 10 Self-Assessment Result
Answer all questions to compute the score.
Frequently Asked Questions
- OWASP Foundation's 2025 list of 10 main security risks for LLM applications with remediation guidance.