During China Cybersecurity Week, Ant AI Safety Lab released SingProbe, an endogenous safety guardrail for large language models that integrates safety detection into the generation process and outputs risk alerts while the model generates. It also released the SingStreamBench benchmark and explored a medical application, SingProbe-Med. SingProbe has been adapted to 29 mainstream open-source LLMs, and its code, models, and benchmark are open-sourced.