July 7

Using AI in Government Websites to Thwart Hackers

Author: Akshaya Ray, CTO, Revize


Mathematician Alan Turing is credited with helping the Allied Forces win World War I by cracking the Nazi encryption machine, Enigma. For most, that’s a pretty cool accomplishment, and Turing could have packed it in and led a retirement of distinguished admiration. But, thankfully for us, Turing was far from finished, and his 1950 publication, “Computing Machinery and Intelligence,” posited something even more incredible than beating Nazis – can machines think? 

Artificial intelligence (AI) is a part of the greater computer science discipline, aimed at answering Turing’s question. AI seeks to simulate and replicate (human) intelligence in machines. The spread has been rapid – self-driving cars, smart-devices to run our lives, and AI-driven marketing, among others. Yet, for all its good, there are still some concerns with AI, and most are security-related. 

Privacy risks that arise from AI are centered on the collection of personal data coupled with the network models that power AI. An astute hacker can poach not only the original data that was used to power the network model but also the multitude of algorithms that spot and store patterns in the data. This has been worrisome for many, and a 2018 Brookings survey confirmed the same, with 49% of respondents believing AI will reduce privacy.  

Yet, in an ironic twist, new research is centered on how AI might aid in mitigating privacy issues. For example, we know hackers are involved in 52% of data breaches. At Revize, we understand that detecting patterns is the best way to prevent future attacks. AI does just this – examines existing behavior and information to then recognize it in advance at a future date. Most hackers rely on similar entry methods, and AI can add uncertainty to their own sequences, which attackers have a hard time adequately predicting.  

Second, AI is still at a point where researchers have shown that by simply placing a Post-it note on a stop sign can fool an AI model into assuming it is now a speed limit sign (due to the square shape). Should a hacker enter looking for a “stop sign,” he or she will encounter a speed limit sign in this instance. By making subtle alterations to existing data, AI changes the algorithms on these neural networks to throw off invaders. 

Revize web and network administrators work around the clock, monitoring logs, security testing, applying the appropriate patches and upgrades, and backups of data and OS. AI holds tremendous potential in our ability to keep our clients safe and secure. We should all be thankful that Professor Turing never drifted into retirement and sipping margaritas on the beach.

CloseClose Form