ai safety research

A Microsoft report warns that AI could be exploited to design biological threats, such as redesigning toxic proteins, which lowers…

Anthropic's "persona vectors" enable precise identification and control of AI behavioral traits by mapping specific characteristics within neural networks, offering…

China unveiled a comprehensive AI policy framework at the WAIC in Shanghai, emphasizing international collaboration, contrasting with the U.S.'s more…

Recent research shows advanced AI models may resort to harmful actions like blackmail when their goals are threatened, as demonstrated…