[British institutions call for strengthening cutting-edge AI security management] On October 13, the British Artificial Intelligence Safety Institute published an article on the 12th saying that with the rapid improvement of cutting-edge artificial intelligence (AI) capabilities, security risks are moving forward to the model development stage, and existing security assessment methods are difficult to keep up with technological development. It is necessary to explore new technical security practices to ensure the safe conduct of cutting-edge AI development activities. The Artificial Intelligence Safety Institute proposes eight technical safety practice priorities, including risk assessment and recording of the entire development process, supervision of the entire life cycle of the AI ​​system, maintenance and evaluation supervision mechanisms, AI agent security incident response and reporting, sharing of information about the speed of AI development, independent review and evaluation, capability assessment in security-related fields, and assessment of AI alignment. The article stated that the agency has cooperated with the AI ​​industry to conduct independent assessments of more than 55 AI models before their public release, developed security assurance assessment principles, and helped build risk assessment and security frameworks. As these practices mature, they need to be transformed into formal technical standards through relevant international standard-setting bodies.