AI2: MIT’s AI Security for Cyber Attacks

Posted on Apr 22 2016 - 10:02am by Devesh

Artificial Intelligence (AI) is everywhere, mostly.  Do you know that your email is being sorted by AI. With the rise of Artificial Intelligence (AI) in most areas, there also arises associated risks/threats. The most prominent of risks/threats are “Cyber Attacks such as DDOS and/or Hacking.” Michigan Institute of Technology (MIT) has been developing technology to counteract this threat. The name of their technology is AI2 (Artificial Intelligence Squared).

We are living in a world where technological advancements are happening within the fraction of a second. With all this technological advancement comes great risks/threats. All our information is mostly online. Are we looking to take care of this at all? Is our online information safe? Are the servers that we post our online information safe and/or secure? Are the servers safe and/or secure from “Cyber Attacks such as DDOS and/or Hacking.” 

Wouldn’t it be great if we were able to predict a cyber attack?  How about doing this before it occurred?  How about preventing and mitigating this cyber attack? Wouldn’t this be the next revolutionary concept? It certainly would be for online information security?

MIT’s security researchers have developed a new AI (Artificial Intelligence). This new AI is based on the next generation cyber security concept. This concept has been called “AI2 or Artificial Intelligence Squared.” The AI2 cyber security concept has the capability for predicting, detecting, and stopping Cyber Attacks such as DDOS and/or Hacking with an accuracy percentage of 85%.

AI2 - MIT's Cyber Attack AI Security

Online or Internet security has become the most challenging in today’s world. There are many victims of cyber attacks such as DDOS and/or Hacking. Most of them are individuals. Corporations and government agencies are also targeted. Hackers and Crackers are finding unique ways threaten the information that needs to be protected. These bad people are able to outsmart the good people. The good people are also not able to keep up.

MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) researchers are working with PatternEx (a machine-learning startup) for developing defenses against these DDOS and/or Hacking threats. MIT’s CSAIL and PatternEx’s team have already developed and deployed the AI2 system. The AI2 system is able to detect 85% attacks after reviewing information from 3.6+ Billion lines of indexed records every day. The AI2 system informs a human analyst if there is something suspicious.

The AI2 system heavily uses Artificial Intelligence (AI). It also relies on human interaction. This capability or capacity is called Analyst Intuition (AI). This two-factor methodology [Artificial Intelligence (AI) and Analyst Intuition (AI)] has given rise to the name AI2 or Artificial Intelligence Squared.

AI2: Inner Workings of The AI Cyber Security System

The AI2 system primarily scans indexed records unsupervised. It uses the machine learning method, and then provides its results to the human analyst at day’s end. The human analyst(s) would then identify all events that are potential and real cyber threats. The human analyst(s) would then also identify all events that aren’t potential and/or real cyber threats. The feedback is incorporated back into the Machine Learning System (MLS). The AI2 system uses this the next time for analyzing new indexed records.

The basic rule is that the AI2 system is able to learn from scanning indexed records. The AI2 system becomes more accurate and better as it continuously analyzes more data such as records and logs. Through testing, the MIT team was able to demonstrate that AI2 was more than 4 times more accurate. The AI2 system was better able to detect automated cyber intrusion attacks. The AI2 system was able to reduce false positives by a significant number.

AI2: Working and Processing Statement

“AI2 continuously generates new models that it can refine in as little as a few hours, meaning it can improve its detection rates significantly and rapidly. The more attacks the system detects, the more analyst feedback it receives, which, in turn, improves the accuracy of future predictions – that human-machine interaction creates a beautiful, cascading effect.”

                                                                                                                                   Nitesh Chawla, CS Professor, Notre Dame University


The AI2 team has presented their working thesis paper, labeled AI2: Training a Big Data Machine to Defend. This working paper was presented at the IEEE International Conference on Big Data Security. The IEEE International Conference on Big Data Security was held at New York City.

Now, we just have to see how AI2 will help create a better and safer Internet. It will be interesting to see how long it takes to implement AI2 on a larger scale. Will most security platforms of the future be integrated with AI2, let’s wait and see.


Disclaimer: The opinions, beliefs and viewpoints expressed by various original authors, other contributors, and other forum participants on this web site do not necessarily reflect the opinions, beliefs and viewpoints of


HASHTAGS:  #AI2 #MIT #AI #Security #CyberAttacks #DDOS #Hacking #MichiganInstituteofTechnology #Cracking #Internet #ArtificialIntelligence #ArtificialIntelligenceSquared #TechnologicalAdvancement #ComputerScience #ArtificialIntelligenceLaboratory #CSAIL #PatternEx #NewYorkCity #NotreDameUniversity #NotreDame #MachineLearningSystem #HumanAnalyst #Analyst #Human #CyberThreat #IEEE #BigData #Security #Indexes #Records #SecurityPlatform #Platform #AnalystIntuition #ArtificialIntelligenceSquared

About the Author

Hi, I am an all-things-tech fan. I have decided to combine my pursuits in #writing and #blogging together to form something #informative for my #readers. I have been associated with the #Blogosphere for the past 9+ years and I love to #browse, #compose, #dissect and #understand new things about #technology and #innovation.

Leave A Response