AI Is Vulnerable to Attack. Can It Ever Be Used Safely?

Simon Makin, Nature

In 2015, computer scientist Ian Goodfellow and his colleagues at Google described what could be artificial intelligence's most famous failure. First, a neural network...