Detection quality

An honest measure of how well TimeProof works: what it catches, what it gets wrong, and what it deliberately leaves out. A case is published only when two different AI models agree; otherwise a human decides.