"Report Says That Action Is Needed to Prevent AI-Based Attacks Winning the Cyber War"
While the use of Artificial Intelligence (AI) in cyberattacks remains relatively limited, a new paper titled "The security threat of AI-enabled cyberattacks" predicts that this will soon change. The paper, a collaboration between WithSecure, the Finnish Transport and Communications Agency (Traficom), and the Finnish National Emergency Supply Agency (NESA), examines current trends and advancements in AI, cyber threats, and places where the two interact. It is noted that AI-based cyberattacks remain uncommon and limited to social engineering applications, such as impersonation, or employed in ways not readily observable by researchers and analysts, such as data analysis in backend systems. However, the paper emphasizes that the number and quality of AI advancements have made more sophisticated cyberattacks possible in the near future. According to the analysis, target identification, social engineering, and impersonation are the most imminent AI-enabled threats today and are predicted to increase in number and sophistication over the next two years. Within the next five years, attackers will likely create AI capable of independently discovering vulnerabilities, planning and executing attack campaigns, leveraging stealth to escape defenses, and collecting data from compromised systems or open-source intelligence. The paper emphasizes that while current defenses can solve some of the difficulties posed by AI-using attackers, others require defenders to adapt and evolve. New strategies are required to combat AI-based phishing that employs synthesized material, biometric authentication system spoofing, and other impending capabilities. The report also discusses the role that non-technical solutions such as intelligence sharing, resource allocation, and security awareness training have in mitigating the threat posed by AI-driven attacks. This article continues to discuss key points made in the new report on AI-enabled cyberattacks.