This paper provides a comprehensive survey of ethical challenges associated with large language models (LLMs), covering both long-standing issues such as data privacy, copyright, and fairness, as well as new-emerging dilemmas like truthfulness and social norms. The authors critically analyze existing research aimed at understanding, examining, and mitigating these ethical risks. They emphasize the importance of integrating ethical standards and societal values into the development of LLMs to guide the creation of responsible and ethically aligned language models. The paper is structured into two main sections: persistent ethical issues and new-emerging ethical issues. In the first section, the authors discuss long-standing ethical problems, including data privacy, copyright, and fairness. They explore the challenges and mitigation strategies for each of these issues, such as differential privacy techniques, backdoor and watermarking methods for copyright protection, and fairness strategies to reduce bias in LLMs. In the second section, the authors focus on new-emerging ethical issues, including truthfulness and social norms. They discuss the challenges of ensuring LLMs generate truthful and accurate information, addressing issues like hallucination and sycophancy. They also explore the ethical implications of LLMs in terms of social norms, including the potential for generating toxic content and the need for alignment techniques to ensure ethical compliance. The paper concludes by highlighting the importance of addressing these ethical challenges in the development and deployment of LLMs, emphasizing the need for a comprehensive approach that integrates ethical standards and societal values into the design and use of these models. The authors also discuss the role of regulatory compliance in ensuring that LLMs are used in a responsible and ethical manner.This paper provides a comprehensive survey of ethical challenges associated with large language models (LLMs), covering both long-standing issues such as data privacy, copyright, and fairness, as well as new-emerging dilemmas like truthfulness and social norms. The authors critically analyze existing research aimed at understanding, examining, and mitigating these ethical risks. They emphasize the importance of integrating ethical standards and societal values into the development of LLMs to guide the creation of responsible and ethically aligned language models. The paper is structured into two main sections: persistent ethical issues and new-emerging ethical issues. In the first section, the authors discuss long-standing ethical problems, including data privacy, copyright, and fairness. They explore the challenges and mitigation strategies for each of these issues, such as differential privacy techniques, backdoor and watermarking methods for copyright protection, and fairness strategies to reduce bias in LLMs. In the second section, the authors focus on new-emerging ethical issues, including truthfulness and social norms. They discuss the challenges of ensuring LLMs generate truthful and accurate information, addressing issues like hallucination and sycophancy. They also explore the ethical implications of LLMs in terms of social norms, including the potential for generating toxic content and the need for alignment techniques to ensure ethical compliance. The paper concludes by highlighting the importance of addressing these ethical challenges in the development and deployment of LLMs, emphasizing the need for a comprehensive approach that integrates ethical standards and societal values into the design and use of these models. The authors also discuss the role of regulatory compliance in ensuring that LLMs are used in a responsible and ethical manner.