Reporting

AI Red Team Report Structure

Executive Summary

  • Number and severity of findings
  • Overall risk assessment
  • Top 3 most critical issues with business impact
  • Key recommendations

Methodology

  • Frameworks used (OWASP LLM Top 10, MITRE ATLAS)
  • Scope and rules of engagement
  • Tools and techniques employed
  • Test duration and coverage

Findings

For each finding:

FieldContent
TitleClear, descriptive name
OWASP LLM IDLLM01-LLM10 classification
MITRE ATLAS IDAML.T0051, etc.
SeverityCritical / High / Medium / Low / Informational
DescriptionWhat the vulnerability is
Reproduction StepsExact prompts, copy-paste reproducible
Proof of ConceptScreenshots, model responses
ImpactWhat an attacker can achieve
Affected ComponentSystem prompt, RAG pipeline, tool integration, etc.
RecommendationSpecific, actionable remediation

Severity Rating Guide

SeverityCriteria
CriticalData exfiltration, unauthorized actions, multi-user impact
HighSystem prompt extraction with credentials, reliable jailbreak
MediumPartial system prompt leak, inconsistent jailbreak
LowInformation disclosure without sensitive data
InformationalTheoretical risk, defense recommendations