Issue link: https://iconnect007.uberflip.com/i/1541367
46 PCB007 MAGAZINE I NOVEMBER 2025 ing ability to act autonomously, adapt to the topic, and focus on the target. While this opens exciting possibilities, deploying agentic AI raises some sig- nificant social and technical questions. As busi- nesses and social groups consider incorporating more advanced AI systems into daily activities, it's important to understand the benefits as well as pos- sible issues and challenges. Advanced agentic AI systems usually include characteristics such as the ability to plan and exe- cute complex sequences of actions under self- direction and adapt to include necessary changes. Some excellent examples include constantly evolv- ing autonomous vehicles, intelligent personal assis- tants that can schedule and reschedule without prompts, or advanced bots that can interact with other systems and humans to achieve specific out- comes, some of which were not anticipated. However, cybercrooks also use agentic AI. Here are some reasons: • Agentic AI can automate complex cyberat- tacks, like phishing and social engineering with little to no human oversight. This means attacks can be launched continuously and repeatedly. • The Agentic AI can learn from previously failed attempts and can adapt on its own to bypass existing security measures. The AI can change communication tactics to avoid detection or modify malware in real time to avoid being caught. • Agentic AI can tailor specific attacks to indi- vidual victims by gathering information and customizing messages or strategies, making scams more convincing and harder to detect and avoid. • It can operate independently. Cybercriminals can seemingly distance themselves from the actual attack, reducing their risk of exposure and prosecution. Agentic AI can perform tasks and optimize much faster than human operators, from breaking pass- words to scanning for vulnerabilities, thus increas- ing the effectiveness and success rate of attacks. Conclusion Agentic AI represents a very powerful evolution in artificial intelligence, providing systems with the ability to act autonomously and adaptively. While this creates exciting possibilities for legitimate use, it also gives cybercriminals a potent tool to success- fully automate and greatly increase their activities, making cyber threats more successful and harder to defend against. As agentic AI develops, under- standing its risks and improving and implement- ing effective defenses becomes ever more critical. PCB007 Dan Feinberg is a technical editor and columnist for I-Connect007. See a list of his columns here.