Automated verification that an AI system's claims are grounded in evidence and don't contain hallucinated or unsupported statements.