The very serious data protection issues aside, today's state-of-the-art biometrics are hardly suitable for authentication purposes (does sample X belong to given person Y?). Using them for large-scale identification (find the most probable person from a huge database, given a sample X), especially with low-quality data sources, is a lottery game.

Unfortunately, the article doesn't state the false-positive and false-negative rates of the technologies used, so it's impossible to calculate the actual risks of being accused of something, once your sample is in the database.