Who Remains in Control During AI-Powered Cyberattacks?
Artificial intelligence is rapidly reshaping the cybersecurity landscape, creating new opportunities as well as new risks. Martina Lindorfer, Associate Professor in the Security and Privacy Research Group, discussed with Eva Komarek how AI is being used in cyberattacks and defence, where new vulnerabilities are emerging, and why human control remains essential. The interview was published by Die Presse as part of its Cybersecurity Week.
In the interview, Martina Lindorfer discusses the growing role of artificial intelligence in cybersecurity and the increasingly professional nature of cybercrime. AI can help attackers create or adapt malicious code, produce more convincing phishing messages, and facilitate social-engineering attacks through technologies such as deepfakes. At the same time, AI and machine-learning methods have long been used in security research to detect attacks, identify anomalies, and automatically classify malware. Lindorfer stresses that AI is particularly effective at filtering and prioritising large volumes of data, but that the final assessment and understanding of an attack should remain with humans.
The discussion also addresses risks associated with smartphones, apps, and the Internet of Things. Long-lived connected devices such as routers, printers, and other IoT systems may remain in use after security support has ended and can provide entry points into larger networks. Lindorfer notes that some vulnerabilities found in IoT devices are based on security problems that have been known for more than a decade.
Looking ahead, Lindorfer identifies two closely related challenges: increasing dependence on AI and the growing complexity of modern systems. She argues that cybersecurity cannot rely on automation alone. Maintaining human expertise, reducing unnecessary system complexity, and considering how technologies can be misused from the outset are essential for building more secure systems. As she emphasises, humans should remain an integral part of any system in which AI is used.
Read or listen to the full interview in Die Presse.
