The traditional legal and technological ethical rules which based on hard law norms have inevitable objective limitations when governing medical artificial intelligence ethics with risk characteristics such as lack of trust in human-computer interaction，uncontrollable decision-making，unreliable security，and unfair distribution of medical resources，which make it impossible to provide the necessary for good governance. Therefore，the dualistic legal model of“hard law soft law”that emphasizes the integrated theoretical structure of“openness -control”can provide a new governance framework for the ethical governance of medical artificial intelligence. Theoretically，the“hard law soft law”paradigm can demonstrate the legitimacy of ethical risk regulation in medical artificial intelligence and ensure the responsiveness of medical artificial intelligence ethical risk decision-making. In practice，the“hard law soft law”paradigm has become a common governance framework for ethical risk management in medical artificial intelligence abroad. In the future，under the paradigm of “hard law soft law”， the ethical governance path of medical artificial intelligence should continuously regulate the generation mechanism of soft law，and enhance the benign interaction mechanism of “hard law soft law”. We also recommend giving full play to the governance advantages of hard law and soft law to achieve institutional complementarity，intending to construct a safe，trustworthy，and responsible ethical governance framework for medical artificial intelligence.