Model denial of service means overloading an AI system with requests or inputs that take a lot of work to process. The system then slows down, stops working or runs up large costs. Common causes are very long inputs that fill the context window, heavy questions sent over and over, and AI agents stuck in a loop that keeps calling the model. Think of a chatbot on your website: someone pastes in a huge document again and again, and every request uses up paid tokens until the bill grows or the service stops answering your real visitors.
Why it matters
Your organization doesn’t need to be a target for this to cost you money. A mistake in an automation can do the same damage as a deliberate attack. If you pay for an AI service by usage, find out whether you can set usage limits and spending caps on your account and on each API key. Ask your vendor what happens when a cap is reached, whether you get an alert before it, and how they stop one user from slowing the service for everyone else.
Published Updated