No matter how great our defences become, it seems that hackers are always a step ahead. However, with new developments in Artificial Intelligence, could security breaches become a thing of the past?
Far beyond the simple-to-detect DDOS, new attack technologies now used against websites can be very difficult to differentiate from normal user activity.
Suspicious usage patterns are becoming ever more stealthy, leaving cybersecurity experts worried that we may be nearing the limits of human ability to predict and uncover attacks before they strike.
Enter AI-Squared, a new platform aimed at monitoring for and identifying cyber attacks. The brainchild of MIT’s CSAIL laboratory and machine learning startup, PatternEx, the technology works by parsing through masses of user-generated data.
It searches for suspicious activity using a neural network that teaches itself the signs of malicious activity, focussing on ‘out of the ordinary’ behavioural cues and user activity. Once the AI flags a target, the results are then sent to a human analyst for further investigation. The security expert’s feedback will then further train the AI, creating better detection.
Surprisingly, this system is estimated to detect roughly 85% of cyberattacks – a number that’s undoubtedly set to rise as technology advances.
Into The Future
AI-Squared, while a big player with impressive statistics, is not without competition: IBM’s famous Watson AI could soon be enlisted for similar tasks.
It’s thought that with the current growth in security Artificial Intelligence related startups and innovations, that we could see an increase in automation and thus decreased need for security professionals.
Will ‘deep learning’ algorithms really replace infosec experts? Time will tell, but for now, security lies well and truly in the hands of our fallible human brains.