A Comprehensive Assessment of Trustworthiness in GPT Models