With the ever evolving technology, there has been an increase in the utilization of IoT, cloud and mobile based applications, resulting in the generation of huge amounts of data. And with this surplus data, we are witnessing a tremendous rise in the criminal fracas that have rerouted to a more unconventional and sophisticated levels than ever before. Incidences such as callousness in segregating password for the high sensitivity information, weak passwords or leaving the access credentials to the hackers’ avail, further adds on to the unfortunate lacunae in cyber security.
Further adding to the shortfall, cyber-attacks have not only gotten more creative and innovative with time but often render its opponents far from chasing the hackers who are in a constant look out to deactivate systems.
And there comes AI:
How can these concerns be addressed? Well, off many things that can be adopted, artificial intelligence is being widely explored to help keep up a pace with this menace. Now, while talking about AI, it’s not necessarily the commonly envisioned sci-fi robots that will come to the rescue of cyber threats but it’s via the implementation through machine learning techniques that AI is doing wonders.
Most of the cyber-attacks in today’s date is figured out by the humans who then quickly block the suspicious communications, thereby keeping a check on their spiteful behaviour. However, by the time people recognize a hack and take actions to combat it, it often gets too late. At other instances, where millions of malicious files are expected to be scanned, the traditional detection methods such as signature-based systems, lack the ability to be truly proactive and hence in a lot of such cases hackers might see an easy opportunity to play the villain.
At this point comes the role of artificial intelligence, a blend of which along with the humans would prove worthwhile in identifying threats. With the machine learning algorithms, computers can gain the ability to learn and make predictions based on previously acknowledged data. It thus implies that AI can react to the newer, unseen cyber threats faster and its ability in recognizing if a file is clean or malicious, is beyond excellence.
AI in cyber security-use case:
Amongst the many developments, the very recent one on the AI based cyber security is the platform called AI2, which is the brainchild of MIT researchers. Put forward by the Computer Science and Artificial Intelligence Laboratory (CSAIL) at MIT, in April 2016, it has the prowess to predict, detect and can stop up to 85% of cyber attacks with high accuracy.
AI2 utilizes machine learning to analyse data and detect suspicious activity. Once that is done, the next step is to feed the data to cyber security experts who give a confirmation if the suspected activity is an attack or a false positive alert. The machine then takes this learning and incorporates it into the next set of data, hence continuously learning and improving.
Developed along with the machine learning startup PatternEx, this platform has analysed more than 3.6 billion log lines, identifying 85 percent of attacks which is three times more effective than the current efficiency that these machine learning platforms hold. That’s not all, the system is also capable of reducing the number of false positives by a factor of five.
In another instance, big data startup Splunk has partnered with consulting firm Booz Allen Hamilton to offer artificially intelligent powered services to help prevent attacks. It would have AI looking for suspicious activities that is not supervised by humans.
While the platforms like AI2 is looked upon highly, the role of artificial intelligence in protecting an organization from cyber attacks and the extent to which trust should be laid upon robots to take care of the security is a topic of debate.
Role of human interventions in these AI platforms, ambiguity in the kind of action that the system would take in case of attack and the concerns over missing out on genuine attacks are some of the major hitches that it possesses.
Though the AI dependent cyber security systems have its own share of glitches, the time is not far away when AI systems would eventually become smart enough to overcome these and avoid instances like false positive results. With the fast learning capabilities AI surely is the future of cyber security, and we hope to see it detecting, preventing and destroying cyber-attacks with much more perfection.
Try deep learning using MATLAB