Aiming at the unsafety and being out of control problems caused by biases in the output of Large Language Model （LLM）， research status， techniques， and limitations related to biases in the existing LLMs were sorted deeply and analyzed from three aspects： bias identification， evaluation， and mitigation. Firstly， three key techniques of LLM were summed up to study the basic reasons of LLMs’ inevitable intrinsic biases. Secondly， three types of biases in LLMs were categorized into linguistic bias， demographic bias， and evaluation bias， and characteristics and causes of the biases were explored. Thirdly， a systematic review of the existing LLM bias evaluation benchmarks was carried out， and the strengths and weaknesses of these general-purpose， language-specific， and task-specific benchmarks were discussed. Finally， current LLM bias mitigation techniques were analyzed in depth from both model bias mitigation and data bias mitigation perspectives， and directions for their future refinement were pointed out. At the same time， the research directions for biases in LLMs were indicated by analysis： multi-cultural attribute evaluation of bias， lightweight bias mitigation techniques， and enhancement of the interpretability of biases.

In view of the rapid development of Large Language Model （LLM） technology， a comprehensive analysis was conducted on its technical application prospects and risk challenges which has great reference value for the development and governance of Artificial General Intelligence （AGI）. Firstly， with representative language models such as Multi-BERT （Multilingual Bidirectional Encoder Representations from Transformer）， GPT （Generative Pre-trained Transformer） and ChatGPT （Chat Generative Pre-trained Transformer） as examples， the development process， key technologies and evaluation systems of LLM were reviewed. Then， a detailed analysis of LLM on technical limitations and security risks was conducted. Finally， suggestions were put forward for technical improvement and policy follow-up of the LLM. The analysis indicates that at a developing status， the current LLMs still produce non-truthful and biased output， lack real-time autonomous learning ability， require huge computing power， highly rely on data quality and quantity， and tend towards monotonous language style. They have security risks related to data privacy， information security， ethics， and other aspects. Their future developments can continue to improve technically， from “large-scale” to “lightweight”， from “single-modal” to “multi-modal”， from “general-purpose” to “vertical”； for real-time follow-up in policy， their applications and developments should be regulated by targeted regulatory measures.