Bridging the Communication Gap: Evaluating AI Labeling Practices for Trustworthy AI Development