As the world is increasingly infused with artificial intelligence (AI), school teachers are beginning to acquire AI literacy and to integrate AI-related content into their teaching practices. However, research on teachers' AI competencies is still in its early stage, leaving many gaps yet to be explored. This study engaged 364 Chinese practicing teachers involved in teaching AI lessons after receiving training, employing a six-factor instrument. The survey assessed teachers' efficacies in understanding AI and teaching AI, with additional considerations of promoting ethical awareness and designing socially beneficial AI applications. In addition, teachers' continuous intention to learn AI and their attitudes toward teaching AI were measured. The survey underwent rigorous validation procedures, confirming its construct validity through confirmatory factor analysis, and demonstrating satisfactory reliabilities and convergent and discriminant validities through other statistical analyses. Structural equation modeling provided support for most of the hypotheses. Further, variance analyses indicated that high school teachers scored higher than primary and middle school teachers across all six measured factors, possibly due to the contextual demands of the university entrance examinations. Overall, the findings suggest a willingness among teachers to enhance their competencies for teaching AI, and underscore the need for increased attention on strengthening teachers' competencies to promote ethical judgement and design AI for social good. The validated survey should be a valid and reliable assessment tool for future teacher development initiatives.