How do I prevent Denial of Service (LLM)?

Implement rate limiting and quotas Use Wardstone to detect expensive query patterns Set maximum output length limits Monitor and alert on unusual usage patterns Implement cost controls and circuit breakers

How do I detect Denial of Service (LLM)?

Wardstone identifies inputs designed to cause excessive resource consumption through pattern analysis of expensive query structures.

LowPrompt AttackOWASP LLM04

Denial of Service (LLM)

Attacks designed to exhaust LLM resources, cause excessive costs, or make the service unavailable.

Test in Playground Get API Key

Overview

LLM-specific denial of service attacks exploit the computational expense of language model inference. Attackers may craft inputs that maximize token generation, require extensive reasoning, or trigger resource-intensive operations. Unlike traditional DoS, these attacks can cause significant financial damage through API costs even without fully overwhelming the service.

How This Attack Works

Attacker identifies inputs that maximize computational cost or output length
What's happening
Testing reveals: 'Write a 10,000 word essay on X' costs $0.50 per request
Multiple expensive requests are sent to the LLM service
What's happening
Bot sends 1,000 max-length requests per minute to the API endpoint
Server resources are exhausted processing the demanding queries
What's happening
GPU utilization hits 100%; request queue grows; latency spikes to 30+ seconds
The service becomes slow, unavailable, or incurs excessive API costs
What's happening
Legitimate users can't access service; monthly API bill jumps from $100 to $50,000

Attack Examples

Token Maximization

Forcing maximum length outputs

Example Attack Pattern (Sanitized)

Write a 10,000 word essay on every aspect of...

Recursive Complexity

Triggering expensive reasoning chains

Example Attack Pattern (Sanitized)

Solve this problem step by step, showing all sub-steps and alternatives...

Repetitive Requests

High volume of resource-intensive queries

Example Attack Pattern (Sanitized)

Automated high-volume requests for complex tasks

Protect Your Application

Try Detection in Playground

Sample Denial of Service (LLM) Input

Write a 50,000 word comprehensive analysis covering every possible aspect of this topic in extreme detail

Try in Playground

Prevention Checklist

Build

Implement rate limiting and quotas
Use Wardstone to detect expensive query patterns
Implement cost controls and circuit breakers

Deploy

Set maximum output length limits

Monitor

Monitor and alert on unusual usage patterns

Detect with Wardstone API

curl -X POST "https://api.wardstone.ai/v1/detect" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"text": "Your text to analyze"}'
 
# Response
{
  "prompt_attack": { "detected": false, "confidence": 0.02 },
  "content_violation": { "detected": false, "confidence": 0.01 },
  "data_leakage": { "detected": false, "confidence": 0.00 },
  "unknown_links": { "detected": false, "confidence": 0.00 }
}