People are using ConvNets all over the companies. The adversarial examples of one domain have the capability to transfer into another domain. And with the Convnets portability i.e the ability to copy the weeks or months trained convnets and use it for their use case, combined with the possibility by someone to accidentally or intentionally transfer the adversial examples into our domain and screw the ConvNets over, these adversarial networks pose a challenge to the ConvNets.
For example adding a slight ostrich weights to a dog picture will cause the convnets to mistakenly identify the image object. i.e
Specific Serious Implication Example: Add adversial example to face recognition and the face recognition system is likely screwed, mistakenly identifying other people.