Will Artificial Intelligence create security immune super-bugs???

For the majority of business leaders, cyber security risks are among their top overall challenges. With machine learning now coming to the forefront of technology it got me thinking about the comparisons to anti-biotics and superbugs that are becoming immune to those drugs. So these "thinking" computers have the ability to deal with the smaller attacks so the human defence can concentrate on the serious threats. This tiering process frees up valuable time and advises us when to get involved. This is all well and good but what happens when hackers are using the same machine learning methods to continually get one up on the security in place?