AI-driven cyberattacks on critical infrastructure are becoming a more immediate threat. Just last November, DP World, a major port operator in Australia, faced a cyberattack that brought operations to a halt for three days, disrupting supply chains across the nation. Now, imagine if the attackers had used AI. Autonomous AI agents could have infiltrated multiple systems at once, bypassing alarms and causing widespread shutdowns in a matter of minutes. While this hasn’t happened yet, experts warn it’s a real possibility. Jamie Moles from ExtraHop puts it simply: “Agentic AI is the next frontier.”
Cybercriminals are increasingly targeting critical infrastructure, often because these systems rely on outdated technology. Alex Yevtushenko from Salvador Technologies highlights this vulnerability, pointing out that such systems are ripe for AI-driven exploits. The risks extend beyond financial losses; they threaten human lives and national security.
But AI isn’t just a tool for attackers. It can also be a powerful ally in defense, enabling real-time threat monitoring and automating security responses. However, AI defenses aren’t perfect. Yevtushenko emphasizes the need for resilience: “Organizations need to focus on resilience—ensuring that even if an attack occurs, they can recover quickly.”
Currently, AI cybersecurity regulations are a patchwork. A unified global framework is crucial to effectively tackle AI-driven threats. Governments should adopt a three-pronged strategy: disrupt cybercriminal activities, encourage AI integration in cybersecurity, and educate operators on using security tools effectively. Moles believes that as AI continues to evolve rapidly, it will be used by both attackers and defenders. “AI will be a key safeguard for security analysts,” he says.
The focus should be on building infrastructure that can bounce back swiftly from attacks. “Critical infrastructure must prioritize cyber resilience for speedy recovery,” Yevtushenko concludes. By taking these steps, we can better prepare for the challenges ahead.
