Verifying AI truth with incentives and rules — not another AI.
Checks AI-generated claims against explicit rules and incentive-weighted verifier scoring instead of a second model's opinion.