# Responsible use notice

R-AI-R v0.5 is intended for authorized defensive evaluation of AI systems, RAG pipelines, document scanners, package-review workflows and agents.

Do not alter the artifacts to request secret exfiltration, unauthorized access, destructive actions, credential use, system-prompt disclosure, real network calls or production actions. Keep evaluator-only files out of the system under test.

The challenge artifacts contain subtle, harmless business and review perturbations designed to test instruction/data confusion. They are not a security certification and do not prove immunity to prompt injection.