E-Handbook: Machine learning security, a real advance in tech protection Article 1 of 4

Don't dismiss machine learning security and what it offers

For the most part, what is often labeled AI in IT security products now is really a subset of AI known as machine learning. That fact -- that when we talk about AI in security products, we're talking about the application of machine learning to security products and not what some security experts considered "true" AI -- disappoints some. But given how hacking has evolved, machine learning security products are becoming essential.

Signature-based detection is increasingly useless for spotting attacks before great damage is done. With the latest hackers able to create and release malware that morphs, identifying known threats is simply not enough. It's the unknown that is going to get you now.

Fighting hackers demands the capabilities machine learning security tools offer, like automation and scalability. They're essential, given the nature of enterprise IT now, with the rapid expansion of mobile users and their interconnected devices. Security products that employ machine learning technology work in a way corporate security teams -- mere mortals -- simply cannot. (Other security experts have pointed out additional advantages, such as machine learning-enabled products' ability to support the zero-trust model of security.) The essential task for security practitioners now is to carefully evaluate offerings, to separate the hype from the effective capabilities.

There've been headlines recently about experts being disappointed that thus far AI is mostly limited to machine learning. Some of those experts -- like Judea Pearl, in his new book The Book of Why -- rightfully argue that AI has thus far not fulfilled its promise. But right now, machine learning security is available and should be seriously considered in any IT team's review of new security products.

Enterprise Desktop
Cloud Computing