XSS from LLM HTML Output

Allow-List Before Downstream Action

Blocking Sensitive Data Exfiltration

Guarding Against Jailbreak Overrides

Sandboxing Model-Generated Code

Sanitizing URLs to Prevent SSRF

Opting Out of Provider Training

Trusting Model Output for Authorization

Masking PII in Model Output

Excessive Agency and Confused Deputy

OWASP Top 10 for LLM Apps

GDPR & HIPAA for LLM Data

Vendor Data Retention & Opt-Out

Injection Risks in RAG Documents

Redact PII Before Embedding

Least-Privilege Design for Tool Access

Trusted Instructions vs Untrusted Data

Over-Privileged Tool Access

Ungated Tool Execution

Logging API Keys in Prompts

Logging Raw Sensitive User Data

Indirect Injection via Retrieved Content

Wrong Role for User Data

Treating Documents as Instructions

Encryption In Transit and At Rest

Trusted vs Untrusted Input Separation

Excessive Agency in Agents

Delimiting Untrusted Content

Data Exfiltration via Outputs

Direct vs Indirect Prompt Injection

Preventing Secret Leakage in Prompts

System Prompt Leakage

Multi-Tenant Data Isolation

Provider Data Retention Policies

RAG Injection via Retrieved Docs

Least-Privilege for LLM Resources

Data Exfiltration via LLM Outputs

System Prompt Leakage and Extraction

Direct vs Indirect Prompt Injection