datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
arabic-guardrail
Arabic Guardrail — 250,842 rows, 12 classes
Defensive dataset for training Arabic prompt-safety classifiers. Each row is an incoming user
message and which of 12 safety classes it belongs to.
بالعربية: مجموعة بيانات عربية لتدريب نماذج تصنّف الرسائل الواردة قبل وصولها للمساعد الذكي.
Arabic guardrails were a gap. Hugging Face searches for Arabic jailbreak / safety /
prompt-injection datasets return zero results, and the one Arabic guardrail model… See the full description on the dataset page: https://huggingface.co/datasets/oddadmix/arabic-guardrail.Evaluation-Dataset-of-AI-Agent-Security-Guardrails
DKnownAI Agent Security Evaluation Dataset
Data Fields
Field
Type
Description
text
string
The adversarial input (prompt) to be evaluated by a security guardrail
action
string
Human-annotated label: blocked or allowed
Citation
@misc{li2026comparativeevaluationaiagent,
title={A Comparative Evaluation of AI Agent Security Guardrails},
author={Qi Li and Jiu Li and Pingtao Wei and Jianjun Xu and Xueyi Wei and Jiwei Shi and Xuan… See the full description on the dataset page: https://huggingface.co/datasets/CaiZhiTech/Evaluation-Dataset-of-AI-Agent-Security-Guardrails.context-adherence-guardrail-10k
Context-Adherence Guardrail — training data (10,710)
Training data for a single-token RAG-groundedness guardrail. Each item is a
(question, context, response) triple with a human- or construction-derived
PASS/FAIL label under one Behavior Spec:
FAIL iff the response makes at least one factual claim unsupported by or
contradicting the retrieved context — truth in the real world is irrelevant
(strict grounding). PASS otherwise, including responses that decline to
answer for lack… See the full description on the dataset page: https://huggingface.co/datasets/aaryand/context-adherence-guardrail-10k.production-ai-guardrail-evals
Production AI Guardrail Evals
Eighteen synthetic, assertion-bearing cases for testing whether a language model can stay inside an advisory role. The cases cover roadmap intake, release readiness, and catalog-change review—the same workload families I use in the Winwood AI Toolkit around IEM Rig.
This is a sanitized public derivative, not a dump of application logs or the private evaluation corpus.
What each row contains
case_id: stable public identifier;… See the full description on the dataset page: https://huggingface.co/datasets/mattwinwood/production-ai-guardrail-evals.guardrail4agent-sft
Guardrail4Agent SFT Dataset
AI 에이전트 Tool Call 데이터 유출 탐지를 위한 한국어 학습 데이터셋입니다.
카테고리
레이블
설명
건수
safe
정상 Tool Call
2,380
S1
PII 유출
1,190
S2
Credential 유출
1,190
S3
Prompt Injection
1,190
S4
내부 데이터 외부 전달
630
S5
시스템 프롬프트 추출
420
분할
data/sft_train.jsonl — 4,900건 (70%)
data/sft_val.jsonl — 1,050건 (15%)
data/sft_test.jsonl — 1,050건 (15%)
synth-health-guardrails
Synth Health Guardrails — consensus-label release
Private release candidate. Licensed under CDLA-Permissive-2.0.
The repository remains private until League Labs publishes the release.
Sensitive-content notice. Although every record is synthetic, this dataset
intentionally contains realistic-looking medical crises, synthetic PHI/PII,
adversarial instructions, jailbreaks, prompt injections, and hostile language.
Synth Health Guardrails contains 13,713 synthetic member… See the full description on the dataset page: https://huggingface.co/datasets/LeagueLabs/synth-health-guardrails.korean_guardrail_testguardrail-arena
guardrail-arena — a two-axis, multilingual guardrail benchmark
from datasets import load_dataset
ds = load_dataset("fevziegeyurtsevenler/guardrail-arena")
Score any LLM prompt-injection guardrail on two axes at once — does it catch attacks
(miss-rate down) and does it avoid over-blocking benign traffic (over-refusal down) — in
English and Turkish. A guardrail is any callable text -> 0|1.
337 prompts: 217 injections (EN 110 + TR 107) · 80 plain benign · 40 security-adjacent… See the full description on the dataset page: https://huggingface.co/datasets/fevziegeyurtsevenler/guardrail-arena.agentic-execution-guardrail-eval
Agentic Execution Guardrail Eval
Agentic Execution Guardrail Eval is a lightweight evaluation dataset for testing safety guardrails in agentic execution environments.
The dataset focuses on risky patterns that may appear when AI agents generate prompts, shell commands, or code intended for execution.
Scope
The evaluation cases cover:
Shell execution and exfiltration patterns
Hidden instructions
Prompt injection attempts
Jailbreak-style reframing
Risky Python code… See the full description on the dataset page: https://huggingface.co/datasets/ole-tail/agentic-execution-guardrail-eval.veto-guardrail-30k
Veto Guardrail 30K
A specialized dataset for training AI security guardrail models to evaluate tool calls against configurable security policies.
Model
This dataset was created to train Veto Warden 4B — a fast, specialized model for real-time tool call security validation.
Overview
Metric
Value
Examples
30,000
Format
Conversational (ShareGPT)
Task
Security policy evaluation
Domains
8 specialized categories
Task Description
The… See the full description on the dataset page: https://huggingface.co/datasets/ycaleb/veto-guardrail-30k.qwen-0.5b_safety_guardrail_dataPlaceboBench
Dataset Card
Dataset Description
PlaceboBench is a hallucination benchmark for retrieval-augmented generation (RAG) in the pharmaceutical domain. It is based on real clinical questions submitted by healthcare professionals to Swedish and Norwegian drug information centers (SVELIC/RELIS), answered by seven state-of-the-art LLMs using retrieved European Medicines Agency (EMA) product information documents as context.
The dataset contains 69 questions spanning 23 drugs, with… See the full description on the dataset page: https://huggingface.co/datasets/blue-guardrails/PlaceboBench.hallucinationThis is a vendored reupload of the Benchmarking Unfaithful Minimal Pairs (BUMP) Dataset available at https://github.com/dataminr-ai/BUMP
The BUMP (Benchmark of Unfaithful Minimal Pairs) dataset stands out as a superior choice for evaluating hallucination detection systems due to its quality and realism. Unlike synthetic datasets such as TruthfulQA, HalluBench, or FaithDial that rely on LLMs to generate hallucinations, BUMP employs human annotators to manually introduce errors into summaries… See the full description on the dataset page: https://huggingface.co/datasets/GuardrailsAI/hallucination.guardrail_samples
Prem Studio Guardrail Datasets
This repo contains two closely related safety/guardrail datasets used in Prem Studio to train small safety models in the style of Llama Guard:
dataset_user_prompt_guardrail.jsonl→ Detect unsafe content in user messages.
dataset_system_response_guardrail.jsonl→ Detect unsafe content in agent/assistant messages (i.e. “did the model reply unsafely?”).
Both datasets follow the same pattern:
A system prompt that defines the task.
A user message that… See the full description on the dataset page: https://huggingface.co/datasets/prem-research/guardrail_samples.ethicalzen-financial-advice-guardrail
EthicalZen Financial Advice Guardrail Dataset
Training data for a safety classifier that detects and blocks financial investment advice while allowing educational financial content.
Dataset Details
Total examples: 407 (204 safe / 203 unsafe)
Train split: 325 examples
Test split: 82 examples
Format: ChatML (system + user + assistant messages)
Labels: JSON structured output with decision, confidence, reason
Task
The model should classify user inputs as:… See the full description on the dataset page: https://huggingface.co/datasets/gayathriethicsai/ethicalzen-financial-advice-guardrail.GRPO-Counseling-Guardrailcompetitor-check
tl;dr:
This is a dataset largely based on CleanCoNLL with some augmentations.
Details:
Base:
We started with the CoNLL-2003 dataset, a standard NER benchmark containing English and German text annotated with four entity types: person, location, organization, and miscellaneous. For our evaluation, we focused solely on examples containing the ORG (organization) entity, as these are most relevant to competitor detection.
We then applied corrections from CleanCoNLL… See the full description on the dataset page: https://huggingface.co/datasets/GuardrailsAI/competitor-check.guardrails-adversarial-banking
Adversarial Banking Guardrails
Dataset Description
This repository prepares and validates a multilingual banking-safety benchmark
for evaluating Large Language Model (LLM) guardrails. The dataset contains
489 adversarial and benign prompts focused on contextual banking fraud,
privacy, fairness, safety, and legitimate finance-related requests.
The checked-in workflow treats the gated source export,
guardrail_banking_combined_context.csv, as immutable. Cleaned v2… See the full description on the dataset page: https://huggingface.co/datasets/sksanjoo23/guardrails-adversarial-banking.preflight-gcp-iam-guardrails
🛡️ GCP IAM Guardrails for Claude Code — free sample
Stop your AI coding agent from running destructive gcloud, gsutil, and bq
commands against your GCP project. This repo is a free sample from the
Preflight studio — catch it before it ships.
Letting Claude Code (or any agent) touch a real cloud project is great until the day it
runs gcloud projects delete or gsutil rm -r on the wrong bucket. These deny rules make
Claude Code refuse those commands outright — no confirmation, no… See the full description on the dataset page: https://huggingface.co/datasets/Cheeisepy/preflight-gcp-iam-guardrails.Counseling-LLM-guardrailGuardrail_test_100
Guardrail_test_100 (EN, KR)
Korean·English Mixed Safety Evaluation Dataset Based on Qwen3GuardTest (Total 200 samples)
Overview
This dataset starts with 100 carefully selected English prompts from Qwen/Qwen3GuardTest, covering 10 risk categories (10 samples per category).These 100 English prompts were then translated into natural Korean using the latest GPT model API, resulting in a total of 200 samples (100 English + 100 Korean).
Main Purposes
Evaluating the safety and… See the full description on the dataset page: https://huggingface.co/datasets/timtkddn/Guardrail_test_100.guardrail-safety-dataset
Guardrail Dataset
Cleaned and deduplicated dataset for training LLM safety guardrails.
Combined from Aegis 2.0, Prompt Injection, and HH-RLHF.
Counseling-Guardrail-Reasoningmisa-ai-2.0-guardrail-ab
misa-ai-2.0 — guardrail A/B (guard on ↔ guard off)
⚠️ Private. Không chuyển sang public khi chưa làm mờ. Cột text của hai subset
internal_eval / internal_eval_dh còn tên lãnh đạo thật, tên cơ quan thật và
prompt nguy hại. Cột system của phần pentest chứa system prompt nội bộ kèm canary.
Muốn public thì phải qua một lượt làm mờ trước.
Đo ngày 2026-09-14 trên test-ai.misa.vn/nlp/api/llm-gateway/v1, model misa-ai-2.0,
harness safety-bench. Hai split khác nhau đúng một biến: system… See the full description on the dataset page: https://huggingface.co/datasets/MISA-core/misa-ai-2.0-guardrail-ab.guardrails-api-test-resultsGuardrail-eval-xstest-beavertail-30k-en-ko
Guardrail Eval: XSTest and BeaverTail 30K (EN/KO)
This dataset bundles two guardrail evaluation benchmarks, XSTest and BeaverTail 30K, into a single Hugging Face dataset repository.
It is organized for multilingual evaluation with English and Korean variants for each benchmark.
What Is Included
xstest_en: English XSTest test set
xstest_ko: Korean XSTest test set
beavertail_30k_en: English BeaverTail 30K test split
beavertail_30k_ko: Korean BeaverTail 30K test split
All… See the full description on the dataset page: https://huggingface.co/datasets/timtkddn/Guardrail-eval-xstest-beavertail-30k-en-ko.