Computer vision systems that measure facial muscle movements and impute emotions are pseudoscience, but people still believe in them. Businesses develop these systems to make money off facial recognition technology. Amazon's recognition API, for example, added fear recognition to its features, deepening the flaws of the original intended use case. This widespread deployment of emotion recognition systems can cause significant harm, especially to communities of color. State authorities should not rely on these systems to inform their actions or decisions because they are built on low-quality labeled data and repackaged race science and physiognomy. The flaws in these systems can lead to errors at different levels and negatively impact people's lives.