Overall, biometric technologies offer many benefits to society, improving convenience, security and commerce. However, biometric technologies are not a monolith. Within biometric technologies for recognition and inference there are component technologies that not only differ in their application areas, but in the computational processes they employ and the data they are trained on. As such, they present unique considerations, benefits, and potential harms, and require distinct policy approaches. Some nascent technologies that require biometric data to function, such as age estimation and augmented and virtual reality, offer unique benefits to protecting children in online spaces, reducing barriers to opportunity, and enhancing equity and inclusion. Because of the scope and scale of biometric data they need to collect for their core functions, some biometric technologies can also present unique risks, such as autonomy and discrimination risks from inferred data about preferences from involuntary or subconscious movements or reactions. The government can help address and mitigate many of these risks by increasing and expanding independent public testing of these systems (as NIST has done for some technologies); developing performance standards for any systems that use biometric information procured by the federal government; and developing more diverse training and evaluation datasets for recognition and inference.