AI Agents Are Testing the Limits of Zero-Trust Security

As autonomous AI agents work together, verifying their actions, enforcing security policies, and maintaining accountability becomes increasingly difficult Vishal Sharma is a Reader in the School of Electronics, Electrical Engineering, and Computer Science at Queen's University Belfast, Northern Ireland, United Kingdom. In a de-authentication attack on an aerial vehicle using a human- and agent-in-the-loop approach, an AI agent can quickly scan the active channel and autonomously intercept traffic, thus reducing the time to react and regain control.