AI & Cybersecurity

The Artificial Intelligence and Cybersecurity working group (Gruppo di Lavoro) is part of the Italian Association for Artificial Intelligence (Associazione Italiana per l’Intelligenza Artificiale, AIxIA), focused on research and innovation on the security and privacy aspects of AI, as well as on applications of AI to cybersecurity. 

As AI becomes more and more pervasive with the introduction, in daily life, of new AI-based applications and services,  a number of security, privacy and ethical concerns have arisen. In fact, as an increasing number of stakeholders look at the AI as an essential tool to process the huge amount of data produced by each environment for commercial uses, governments and legal authorities are starting to look at the implications related to privacy, physical and software security, and human ethic which are brought by usage of AI-based services and applications.  

Artificial Intelligence is a powerful tool for cybersecurity practitioners, effectively used to identify, tackle and prevent cyberattacks, identifying malware ahead of time, and detecting anomalous behaviors. Still, in recent years, cyberattackers are also exploiting AI-powered mechanisms to carry more efficient and effective targeted cyberattacks. From another angle, AI can be used to learn behavioral features of software, systems or human users, becoming thus a main enabler for key security processes such as identification, authentication and authorization. At the same time, AI-based technologies have been extensively used by intelligence agencies and repressive governments to systematically violate citizen privacy and track at the same time movements and private life of both criminals and dissidents. Thus, AI is a powerful and flexible technology, which has become an essential tool in fighting cybercrimes. Still, it can be used for malicious or exploitative purposes targeting systems integrity, availability and people’s privacy.  This raised several concerns on the possible misuses of AI-based technologies, recently pushing for establishing a regulatory framework, such as the one recently proposed by the European Commission, to classify AI applications based on their risk level and regulating their possible usage for each level. 

From these considerations, the AI & Cybersecurity Working Group aims at promoting research and practice on two main wide-topics where AI and Cybersecurity are intertwined:  

i) AI-enabled/empowered cybersecurity applications and services (Cybersec through AI); 

ii) Security and privacy aware AI, according to the concepts of trustworthy artificial intelligence and human centered AI (Cybersec for AI).  

The following non-exhaustive list reports the topics of the Working Group: 

• Trustworthy AI

• Privacy Preserving Data Analysis

• Explainable AI

• AI powered cyberattacks

• AI based malware analysis

• Resilient and Robust AI

• Human Centric and Ethical AI

• Behavioral Analysis

• Intrusion Detection

• Resilience and Robustness of AI

• Adversarial attacks

• Misbehavior detection through AI-based multimedia analysis

• Threat Prediction

• Legal and ethical aspects of AI and its applications

• AI-based policy management

• Privacy aware AI training

• Dataset Protection

• Secure Federated Learning

• Biometric Recognition

Andrea Saracino, Istituto di Informatica e Telematica, Consiglio Nazionale delle Ricerche

Francesco Mercaldo, Università degli Studi del Molise

Tamara Devalle, Protiviti