United in the name of cyberdefense

, by Greta Nasi - direttrice del MSc in Cyber risk strategy and corporate governance
New technologies based on machine learning allow advantages in timeconsuming activities such as the analysis of great quantities of logs. Since they employ vast amounts of data, however, it is absolutely necessary to ensure their security and privacy

Cybersecurity requires governing several processes, including scanning, log analysis, incident response, and threat detection. These processes can be time-consuming and labor-intensive, demanding security analysts spend significant time reviewing and analyzing data. Automation can help streamline these processes by reducing the need for manual intervention, allowing security teams to respond more quickly and efficiently to security incidents. As Artificial Intelligence (AI) technologies become more prevalent and powerful and drive digital transformation through automated decision-making capabilities, their application is rapidly growing.

While the benefits of this emerging technology are significant, there are also concerns to be aware of. AI techniques and systems may produce unexpected outcomes and be susceptible to tampering that could manipulate expected outcomes. Consequently, cybersecurity and artificial intelligence are increasingly intertwined. On the one hand, AI can enhance cybersecurity by identifying and mitigating threats more effectively. On the other hand, AI itself can be vulnerable to cyber-attacks which could compromise data and systems (Li, 2018).

The advantages of AI in cybersecurity are numerous (Floridi and Taddeo, 2018). AI can aid in automating tasks, identifying abnormalities in real-time, and responding to cyber threats with greater speed and efficiency. One critical process in cybersecurity is threat detection and response, which involves identifying and responding to potential cyber threats and attacks. The goal is to detect suspicious activity or anomalous behavior on a computer network, system, or application and respond quickly to mitigate the risk, minimise possible damage and prevent further harm. It involves repetitive and continuous monitoring of systems and networks to identify unusual behavior that threatens business continuity. Advanced threat detection technologies include machine learning and AI algorithms. These can help automate threat detection and behavioral analysis activities and be trained to identify patterns and anomalies in network traffic, responding to cyber threats accurately and quickly.

AI technologies are also currently employed in vulnerability management to identify technical risks in hardware and software systems allowing organizations to patch and update them before attackers can exploit them. Organisations employing complex computer networks as critical infrastructure operator that governs a power plant can significantly benefit from these technologies, as manual vulnerability scanning would not be feasible.

Overall, AI can potentially transform cybersecurity operations in the future. Thus, its implementation has challenges. They include the need for more skilled professionals, the importance of solid organizational data governance, and the potential for AI to be breached or manipulated. AI technologies offer many benefits in improving cybersecurity and other tasks and processes that typically require human intelligence in terms of automation, improved decision-making, personalized experience based on the users' preferences and increased safety and security in several areas such as food policy, transportation, and healthcare. However, they also pose some security challenges.

Firstly, AI technologies rely heavily on data to learn and make decisions. The security and privacy of data are critical (Zhang et al, 2019). Suppose sensitive data is compromised or accessed by unauthorized parties. In that case, it can be used to launch data breaches that may serve several purposes as cyber-attacks, industrial espionage, the launch of misinformation campaigns and other security incidents. In addition, it may be hard to understand how some AI systems make decisions. This needs more transparency to ensure identifying potential privacy and security risks.

Secondly, AI models can be biased, leading to unintended or even discriminatory outcomes. Bias and fairness challenges may lead AI algorithms to make incorrect decisions, which become malicious in the case of adversarial attacks that target the AI models to manipulate the data they use to learn and make decisions.

To address these challenges, it is essential to implement rigorous privacy and security measures to manage data properly and ensure the fairness and transparency of AI systems while providing human oversight and intervention. All organizations using AI technologies should adopt appropriate security policies and procedures specific to AI, invest in security training and awareness for AI professionals, and regularly audit and test their AI systems to identify vulnerabilities and address potential security risks.

As AI and cybersecurity rapidly develop, their integration is expected to offer increasing application scenarios. Future developments may include the study of AI-based situational awareness for cybersecurity, which can provide intelligent prediction and protection for cyberspace; the development of novel and specialized AI algorithms for cybersecurity, particularly for big data intelligence; and innovative security protection solutions for AI in the future.