Like many people, you have probably come into contact with facial recognition technology. In some cases, you may have been aware of it, such as if you use it to unlock your phone. In others, such as when passing through airport security, you may have been unaware it was in use. Facial recognition technology is also used by Colorado law enforcement, but it is not always as accurate as it may seem at first.
Disparities according to race
One potential issue with facial recognition technology is that it may be more likely to misidentify individuals with darker skin. Testing by researchers and by the National Institute of Standards and Technology, a federal agency, has found ongoing issues in this area. This could have implications for a sucessful criminal defense if a darker-skinned person is identified using facial recognition technology.
Real world usage
Another issue is that when companies report the accuracy of their facial recognition technology, those results are not necessarily transferable to how the technology will be used in real-world situations. For example, police may be working with images that are poor quality, or their database of images may be of the wrong size to achieve the same kind of results. Companies may report a certain rate of successful identification or of low false positives based on their own testing, but these might be in controlled conditions that do not reflect the conditions in which law enforcement uses them.
These misleading performance scores that companies claim for their facial recognition software can make a positive identification in a criminal case seem more likely to be accurate than it really is. Concern about the technology’s reliability as well as about privacy has led organizations such as the American Civil Liberties Union and some lawmakers to press for legislation that prohibits its use by law enforcement.
