A new “red team” will try to anticipate and thwart attacks on machine learning programs.
http://www.wired.com/