AI Can Be an Ally Against Terrorism
OPINION |

AI Can Be an Ally Against Terrorism

INVESTIGATING THE DARK WEB BY EMPLOYING MACHINE LEARNING CAN BRING TO LIGHT THE CORRELATIONS BETWEEN FUNDING, RECRUITMENT, AND ATTACKS

by Nicola Limodio, Dept. of Finance, Bocconi


The Revamp of Terrorism
The last decade experienced a surge in the number of terrorist attacks. Although the reasons behind this wave span greatly from religious intolerance to separatist desires, the outcome of such attacks are unsurprisingly regular: a culture of fear and insecurity, emotional and human costs and financial markets depressed by increased risk.
 
Countless organizations, both national and international, have sworn loyalty to the battle against terrorism by investing to understand its causes and implications. Although some consensus exists regarding the determinants contributing to the birth of pockets of unrest within a country, much still needs to be unveiled regarding the mechanics through which extremist groups operate. Understanding the channels enabling terrorists to function boost the chances of stopping their doing. Machine learning and artificial intelligence may be able to provide a groundbreaking set of new tools to this battle.
 
Information Communication Technology and Terrorism: a Match Made in the Dark
How are terrorists able to finance attacks, organize operations and recruit people? This is a fundamental question, given that as well-known attacks like 9/11 and Bataclan showed, extremist groups articulate their activities through multiple channels and operational devices. As the literature in this field highlighted, terrorist organizations are structured through sophisticated layers of hierarchy and rely on monetary and career incentives beyond religion and the after-life.
 
This research makes use of a particular media, which permits to understand and study the behavior of terrorist groups: Jihadi-friendly platforms operating in the dark-web. Considering what is going on in those internet domains may represent a true breakthrough in the fight against terrorism.
 
These platforms are present in non-publically available part of the web, accessible through a specific piece of software called “The Onion Router” (TOR). They specialize in the exchange of messages on various topics, from commenting news to exchanging information on local events, and also coordinating individual activities and recruiting new terrorists.
 
 
Machine Learning and Recruitment Activity on The Dark Web
How can we make use of the information present on dark web to counter terrorism?
Machine Learning is the science of getting computers to learn and act like humans do, and improve their learning over time in autonomous fashion, by feeding them data and information in the form of observations and real-world interactions.
 
In a recent study, I use information from 28 online message boards in 5 languages and study the content of these platforms by focusing on the timing of the messages, the wording and intents.
 
By using machine learning on data extracted from such message boards, I introduce algorithms that are able to detect messages encouraging terrorist actions or recruiting followers. In order to implement such method, I scraped data from these platforms operating in the dark web and collaborated with judges, by offering a random sample for evaluation. Such judges read, evaluated and categorized which messages were aimed at recruiting terrorist. After this verdict, I have fed the results to a machine learning algorithm and trained it to identifying the recruitment messages among the remaining 17 million un-categorized messages.
 
 
Recruitment, Financing and Attacks – A Path for the Future
These results permitted to define periods of intense terrorist recruitment as those in which these platforms host a large share of recruitment messages. Such measure was then coupled with a specific setting, in a developing country, in which I could join information on shocks to the funding of terrorist groups. A key result of this research showed that there exists a strong complementarity in the number of terrorist attacks: these are particularly numerous and deadly when both increased financing and intense recruitment take place.
 
As of now, such intersection between finance, data science and national security is still in an embryonal phase, but it represents a precious way to use virtual domains to unpack the functioning of terrorist organizations, understand their operations and provide sound counter-terrorism advice. I plan to extend these results over the next months and exploit this innovative methodology to research the causes of terrorism and the rise of radicalization.
 
 

Latest Articles Opinion

Go to archive
  • With Increasingly Flexible Work, the Old Stakes Are of Little Use

    New work time arrangements and forms require new ways of management and new protections. And therefore, new forms of labor relations and union organizations

  • Trust That Can Be Trusted

    There is social trust and institutional trust. And if the former is more stable to external shocks, the latter is more sensitive to them. The two can also diverge, as during the pandemic, when mistrust in US institutions towards emergency management corresponded to an increase in social trust

  • Reskilling Is the Key

    According to a survey by the World Economic Forum around the world, in 2027 42% of jobs will involve automation. The competitiveness of workers will therefore increasingly depend on training that brings them up to date with the new needs of the market. For this to happen, however, they need to grasp its usefulness

Browse the magazine in digital format.

View previous issues of Via Sarfatti 25

BROWSE THE MAGAZINE

Events

Mon Tue Wed Thu Fri Sat Sun
        1 2 3
4 5 6 7 8 9 10
11 12 13 14 15 16 17
18 19 20 21 22 23 24
25 26 27 28 29 30 31