Denial of Wallet
An attack that intentionally increases AI usage or inference costs by triggering excessive requests, expensive operations, or unnecessary model executions.
What is Denial of Wallet?
Denial of Wallet attacks exploit the usage-based costs of AI and cloud services. An attacker may repeatedly send expensive prompts, trigger resource-intensive model operations, create unusually long interactions, or automate large numbers of requests. Unlike traditional denial-of-service attacks that primarily target availability, DoW attacks can focus on exhausting financial or computational resources.
Why is Denial of Wallet Important?
Generative AI applications often incur costs based on tokens, API requests, compute usage, or other resources. Uncontrolled consumption can result in unexpected expenses, degraded performance, and service disruption. Rate limits, usage monitoring, access controls, and spending thresholds can help reduce this risk.
Common use cases
Denial of Wallet protections are relevant to LLM APIs, AI agents, generative AI applications, cloud AI services, public chatbots, and usage-based AI platforms.