Abstract: Constructing the new-type power system integrated with an increasing proportion of renewable generation is an important measure to achieve the goals of "carbon emission peak and carbon neutrality." The high uncertainty of renewable power and the weak immunity of power electronic equipment bring a bigger security risk to the operation of the new-type power system. Security risk early warning performs security analysis for future power system operating conditions (OCs), identifying the security risk and guiding control decision-making. It is an effective technique to safeguard the secure and stable operation of the new-type power system. To improve the computing speed of security risk early warning, data-driven security risk early warning methods draw more attention. First, the new features of the source side, network side, load side, and energy storage of the new-type power system are elaborated, which include the integration of large-scale renewable generation, the interaction of ultra-high-voltage AC and DC systems, the integration of distributed renewable generation and new-type loads, and the application of energy storage for power regulation and control. Then, the main dynamic security risk of the new-type power system is introduced, and the security risk early warning framework is formulated. Next, the domestic and foreign research work of four main parts is reviewed, which include the generation and reduction of OCs, contingency screening, security risk assessment, and insecure severity grading. Finally, with the characteristics of the new-type power system considered, the problems in existing data-driven security risk early warning research work are analyzed, and future research directions are prospected.