The company is accused of illegally scraping 20 billion images of people's faces from the web without their knowledge or permission, and then using them to form a global facial recognition database.