Can't understand how a "very" secure app considers this an acceptable part of the threat model.