With the rapid development and continuous integration of artificial intelligence into various fields，computing power has become one of the key factors influencing its development and application. In recent years，significant progress has been made in the construction of artificial intelligence computing infrastructure. However，due to the diverse nature of the infrastructure，complex nodes，and the inherent vulnerabilities of artificial intelligence itself，security issues such as data and model theft，adversarial attacks，and untrusted nodes have emerged，affecting user trust. Artificial intelligence computing infrastructure serves as both “infrastructure” and “artificial intelligence computing power，” and it is also a “public facility”. It possesses the triple attributes of infrastructure，technological aspects，and public elements. Accordingly，promoting the secure development of artificial intelligence computing infrastructure should focus on three aspects：strengthening its own security，ensuring operational safety，and facilitating security compliance. This can be achieved by enhancing reliability，availability，and stability，safeguarding the confidentiality and integrity of algorithms during runtime，and improving users’ security control，acceptance，and compliance in eight domains. These efforts aim to fortify the defense line of artificial intelligence security，build a trustworthy，usable，and user-friendly foundation for artificial intelligence computing power，and create a secure，healthy，and compliant development ecosystem for the artificial intelligence industry.