Definition
This is when an attacker overwhelms or crashes an AI agent's software so it stops responding to legitimate users, similar to flooding a phone line so no real calls get through. Because AI agents often run continuously to power chatbots, dev tools, or IoT devices, even simple flaws in how they process requests can be abused to take them offline.
Why it matters
Any business process that depends on an always-on AI agent (customer support, automated trading, device control) can be disrupted with minimal attacker effort, causing operational and reputational damage.