Generative artificial intelligence, represented by ChatGPT and DeepSeek, refers to artificial intelligence that can generate text, pictures, videos, and other corresponding content according to user instructions. Personal information is the basis of generative artificial intelligence. Generative artificial intelligence needs to deal with a large amount of personal information in all stages of model training, model generation and model optimization, which also brings a certain impact on the traditional personal information protection rules. In the stage of information collection, generative artificial intelligence may blur the rules of informed consent and infringe the privacy of information subjects. In the stage of information utilization, generative artificial intelligence may impact the basic personal information processing rules such as the principle of purpose limitation and the principle of openness and transparency, and increase the risk of personal information disclosure. In the stage of information generation, generative artificial intelligence may generate false information and discriminatory information. In the context of the transformational development of generative artificial intelligence, it is urgent to examine the basic concept of personal information protection and seek its value orientation in the field of generative artificial intelligence. By examining the development of comparative law and the concept of personal information protection in China, we can see that the unipolar thinking of personal information protection or personal information utilization is difficult to adapt to the practical needs of the digital society, and the dynamic balance between personal information protection and personal information utilization is the ideal path to properly balance the interests of various subjects. Generative artificial intelligence can be widely used as a basic model in many fields such as education, finance, science, and technology. In view of this, a balance between personal information protection and generative artificial intelligence development should be coordinated and promoted. Personal information is closely related to the information subject. Once personal information is disclosed or abused, the information subject may suffer higher risks. Therefore, it is necessary to build a collaborative relief mechanism of risk prevention and damage compensation, to promote the benign development of generative artificial intelligence based on whole life cycle protection of personal information. As far as the risk prevention mechanism is concerned, it is necessary to improve the de-identification measures based on risk identification, and grant the information subject the right to limit processing and the right to interpret algorithms, to curb the potential risks in an all-round way. Determination of the subject of liability is the basis of damage compensation. It should be proved by the service provider of generative artificial intelligence and the user that there is no causal relationship between them and the damage, otherwise they need to bear joint and several liability for compensation. In terms of the principle of imputation, the principle of presumptive fault liability or no-fault liability can be applied respectively according to the fact that the infringed object is personal general information or personal sensitive information. In order to better remedy the damage suffered by the information subject, in addition to the traditional compensatory compensation such as property damage compensation and mental damage compensation, punitive compensation should also be introduced to protect the damaged rights and interests of the information subject to the greatest extent.