Unbounded Consumption in AI risk refers to?

• A. End Users Drawing Excessive Inferences, Causing DoS Attacks Or Financial Loss
• B. End Users Consuming Unlimited Storage
• C. The Model Producing Infinite Outputs
• D. The System Failing Due To Memory Leak

Answer: (A)

Unbounded consumption is about how users or automated agents push AI services beyond reasonable limits, extracting more value and inferences than intended and consuming resources aggressively. When end users draw excessive inferences from outputs or run large volumes of requests, the system can become overloaded, leading to denial of service for others or triggering higher costs and potential financial losses. This taps into the real risk of abuse or overuse of AI capabilities, where the demand on compute, bandwidth, and money spirals beyond what the service is designed to handle.

The other ideas don’t fit this risk pattern. Unlimited storage is a capacity issue, not about how users interact with the AI service. Infinite outputs would imply the model itself producing endless results, which is a different failure mode. A memory leak points to a software bug, not to usage behavior by end users.