Red-teaming Baichuan 2 LLMs, three new LLM safety benchmarks, Foreign Ministry on global AI governance, and new standard on algorithm security