[mmooss]

Using AI systems for high-speed security actions, proactive and reactive, seems necessary but not sufficient:

I expect attackers will also use AI systems, trained on the latest in effective attacks. What about defense would make defenders' AI systems more effective than attackers'?

I think it's necessary because, if the attackers use AI systems then the defenders need to keep up.

Also, we need to be creating far more secure systems to start with. Now it is, to a degree, security through obscurity - something is secure when attackers can't find the bugs fast enough. Security through obsurity wouldn't seem to work well when the attacker uses AI software.