Is AI cybersecurity the next big tech leap?

Deep learning is a useful tool to optimise and validate security posture. But until we overcome some of its challenges, positive security models and behavioural algorithms that are deterministic and predictable are still more effective for defence and mitigation. Most successful deep-learning applications in use today are based on supervised learning neural nets. They take an input and produce an output where the output provides a confidence level across a fixed set of labels. Given lots of data, the neural net will usually make the right "decision".