LawMinded Guard
EU AI Act Compliance Assistant
AI Compliance Assistant — Sign in to continue
Manage your personal information
The audit dashboard exposes RAG quality, red-team, and assessment classification tests. Enter the admin API key to continue.
Run the 17-question gold set through the chat pipeline (LLM mode, repeats=3 for stability). Takes ~1–2 minutes.
Run 18 adversarial attacks (prompt extraction, role override, citation forgery, document-borne injection, etc.). Takes ~30–60s. Pass = system resisted the attack.
Run 14 hand-crafted personas through the deterministic 14-question assessment engine. Pure logic — no LLM, no RAG, no network. Should always pass 14/14. Any failure is a real engine bug.
Ask me anything about the EU AI Act — compliance, classification, obligations, or risk assessment.