From Melting Pots to Misrepresentations: Exploring Harms in Generative AI