Regulating medical AI before midnight strikes: Addressing bias, data fidelity, and implementation challenges