Team Ai
Datasetpublic

zachz/prompt-injection-benchmark

Prompt Injection Benchmark A curated dataset of labeled prompt injection attacks and benign prompts for testing and benchmarking injection detection systems. Dataset Description This dataset contains 200 examples across 7 attack categories, plus 100 benign prompts. Each example is labeled with: text: The prompt text label: injection or benign category: Attack category (e.g., instruction_override, role_hijack) severity: low, medium, high, or critical… See the full description on the dataset page: https://huggingface.co/datasets/zachz/prompt-injection-benchmark.

sourceHugging Facemitupdated 6mo agoView on Hugging Face
1likes260downloads
Dataset Card

Prompt Injection Benchmark

A curated dataset of labeled prompt injection attacks and benign prompts for testing and benchmarking injection detection systems.

Dataset Description

This dataset contains 200 examples across 7 attack categories, plus 100 benign prompts. Each example is labeled with:

  • —text: The prompt text
  • —label: injection or benign
  • —category: Attack category (e.g., instruction_override, role_hijack)
  • —severity: low, medium, high, or critical

Attack Categories

CategoryCountDescription
instruction_override30"Ignore previous instructions" variants
role_hijack30"You are now..." identity takeover
systempromptleak25Attempts to extract system prompts
delimiter_injection25Fake system/assistant markers
encoding_bypass20Base64, Unicode trick attacks
jailbreak35DAN, safety bypass attempts
data_exfiltration35Extract data or make external requests
benign100Normal, safe prompts

Usage

python
from datasets import load_dataset
ds = load_dataset("zachz/prompt-injection-benchmark")

Use Cases

  • —Benchmark prompt injection detectors
  • —Train classifiers for injection detection
  • —Red-team testing for LLM applications
  • —Security auditing of AI systems

License

MIT