Do Generative AI Models Output Harm while Representing Non-Western Cultures: Evidence from A Community-Centered Approach