AI-Automated Cybersecurity: What to Automate?

Picture by Editor

Let’s face it: Whereas some IT professionals could have a knee-jerk response in opposition to AI due to the present hype, it’s only a matter of time earlier than AI turns into embedded into many every day enterprise processes, together with cybersecurity controls. However now, when this know-how continues to be younger, it may be obscure the true implications and challenges of AI automation.

This text debunks a few widespread myths about how AI can improve cybersecurity and supplies IT and cybersecurity leaders with suggestions on methods to make knowledgeable selections about what to automate. 

Don’t purchase into the parable that AI goes to switch all of your staff. Even when that have been attainable, we as a society usually are not prepared for that leap. Think about boarding a jet and noticing that no human pilot ever enters the cockpit previous to departure. Little question there can be mutiny on board, with passengers demanding {that a} pilot be current for the flight. As efficient because the autopilot operate is, it has its limitations, so individuals nonetheless need a human in cost. 

Certainly, we didn’t see human personnel purged again when the economic revolution took maintain. Whereas equipment did take over parts of guide labor, it didn’t change the people themselves. Relatively, the machines introduced better effectivity, predictability and consistency to the manufacturing course of. In actual fact, new jobs and even new industries requiring new abilities and better variety have been born. Equally, AI will convey new ranges of effectivity, scalability and accuracy to enterprise processes, and likewise create new alternatives and rework the labor market. In different phrases, you’ll nonetheless want cybersecurity personnel, however they are going to be upskilled by AI help. 

One other vital false impression is that AI automation will inevitably scale back prices. This may occasionally sound acquainted; the identical was mentioned concerning the cloud not too way back. Organizations that migrated their datacenters to the cloud discovered that whereas the OPEX value construction of the cloud has benefits over conventional CAPEX expenditures, the ultimate prices are comparable for giant environments, partly as a result of extra refined methods require extra expert (and costly!) expertise. Likewise, automation will change the distribution of prices, however not the general prices. 

Lastly, a totally automated AI-driven safety answer is typically seen as a fascinating aim. In actuality, it’s a pie-in-the-sky dream that raises questions of belief and auditability. What if that automation malfunctions or turns into compromised? How do you confirm the outcomes are nonetheless aligned with the enterprise aims? The reality is that we’re within the early levels of this new AI automated paradigm, and nobody really understands how AI automation is likely to be exploited sooner or later from a safety perspective. AI and automation aren’t silver bullets (nothing is).

Sure processes are higher fitted to automation than others. Right here is an effective three-point evaluation that may enable you to resolve whether or not a safety course of is appropriate for automation:

• The method is repetitive and time consuming when carried out manually. 
• The method is sufficiently effectively outlined that it may be became an algorithm.
• The outcomes of the method are verifiable, so a human can decide when one thing is incorrect.

You don’t need your costly safety expertise doing issues like pouring over safety logs, correcting misconfigurations or decoding prescribed metric alerts. By equipping them with AI-driven safety instruments, you possibly can enhance their visibility, increase their understanding of various threats and expedite their responsiveness to assaults. 

Extra broadly, think about how skilled sports activities groups are investing in know-how to enhance the efficiency of their athletes. Equally, you’ll want to present your safety groups with the automated instruments they should up their sport. For instance, the insider menace is a big danger, however it’s virtually not possible to look at over each consumer within the firm, and rogue staff are sometimes solely evident after they have already prompted a minimum of some harm. AI-based options will be rather more environment friendly in decreasing this danger: A consumer and entity conduct anomaly (UEBA) detection answer can spot delicate modifications in a consumer’s information entry patterns and variations between their conduct in comparison with their friends, each of which sign a possible danger that requires immediate evaluation. 

One other space the place AI can take your crew’s capabilities to a complete new degree is menace searching. Automated options can determine extra precisely traces of assaults that will have been thwarted by your safety mechanisms and examine them to your menace intelligence. These could also be indicators of a bigger assault and you will get higher ready for it. 

ChatGPT, Bard and 1000’s of different wonderful new apps give executives the chance to expertise AI in motion. Working with their safety groups, they will discover potential functions for the know-how. However as an alternative of blindly charging ahead, it’s important to completely assess which processes make sense to automate. This due diligence will assist IT leaders be certain that the dangers of a proposed new know-how doesn’t exceed its advantages.
 
 

Ilia Sotnikov is Safety Strategist & Vice President of Person Expertise at Netwrix. He’s liable for technical enablement, UX design, and product imaginative and prescient and technique.