Taming AI Hallucinations: Galileo Labs’ New Metrics for Safer and More Reliable AI