There is a patent claim by Christian Szegedy and Ian Goodfellow for the adversarial training of neural networks:
I am not sure what implications this has. Does this mean that any company doing adversarial training of their neural networks must now pay loyalties? What about academics?
Source: Reddit Machine Learning