Italy needs to build IT protection tools based on artificial intelligence on its own: in a context where “almost all the information and automation systems of all sectors of the country use foreign hardware and software, we must at least have the ability to control what happens in them, using tools that remain under our control ”.
To say it is Nicola Mugnato, CEO and co-founder of Gyala, a 100% Italian startup founded in 2017 and today thanks to the know-how gained by developing IT protection projects for critical infrastructures with the Ministry of Defense and with the main national System Integrators. A somewhat anomalous startup, because it is not made up of twenty-year-old graduates, but of “twenty-year-olds with thirty years of experience who use management and industrial development techniques typical of multinationals”. In particular, the company realizes all-in-one platforms able to automatically prevent, identify and manage cyber threats and anomalies 24 hours a day. Threats that are among other things in continuous growth: according to the latest Clusit report, last year alone there was an increase in cyber attacks of 12% on a global scale, confirming a trend that has been growing for at least 4 years.
The fault is also of the pandemic: lockdown and closures have imposed a hasty (and often problematic) digitization of companies at all levels, even forcing them to set up smart working for an unprecedented number of people, and this has in fact greatly enlarged their perimeter of vulnerability computer technology; Furthermore, protect an employee from cyber attacks when she works from home it is more complicated than when she is in the office, sheltered within a corporate network.
What companies risk
“There are essentially two types of risks for companies – Mugnato explained to Italian Tech – random and targeted attacks. All of us and small and medium-sized enterprises are mainly exposed to the former, while large companies, the Public Administration and national critical infrastructures both have to face these threats ”.
The random attacks they are the ones we expose ourselves to when we use the Internet to receive emails or to browse because, through these services, “we can access compromised or fake sites, created to get hold of our username and password – clarified the CEO of Gyala – Or for let us download malware or ramsonware, who can encrypt our data by taking them hostage and then ask for a ransom to unlock them “. There is no specific goal: they are traps set off waiting for someone to inadvertently step on them.
The targeted attacks are quite another thing: designed and built to target a specific organization or company, they can have various objectives, “from creating an image damage or propagating a certain idea, as in the case of Hacktivism, up to industrial espionage and even terrorism, when it strikes to alter or interrupt an essential service such as the supply of energy or the chlorination of water ”, Nicola Mugnato told us.
cinema and tv
The 10 best movies (and series) about hackers
by Riccardo Meggiato
How AI can help us
Given this context, it may be comforting to know that artificial intelligence contributes more and more to counteract all types of attacks: “The most used algorithms are those of anomaly detection, which implement dynamic analysis of a system to search for unusual behaviors that are difficult to identify by a human operator ”, Mugnato revealed to us. There are two types of data useful for revealing a possible cyber attack: communications between computers and the programs that run on them. The algorithms of AI, therefore, “are used to understand both if a program is good or bad and when a good program is used in a bad way”.
Finally, in managing threats through systems based on artificial intelligence, like those developed by Gyala (whose platform is called Agger), there are two main moments: the identification of the threat and the reaction. In the first, AI algorithms are used that self-learn thanks to machine learning, while in the reaction phase “we prefer to use the so-called expert systems, where the skills of specialized human operators have been transferred”, ie machines that they reproduce the behavior of an expert in flesh and blood making similar choices, but with greater speed and efficiency. At least, that’s the intention.