datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Nemotron-Content-Safety-Audio-Dataset
Nemotron Content Safety Audio Dataset
Dataset Description
The Nemotron Content Safety Audio Dataset is a multimodal extension of the Nemotron Content Safety Dataset V2 (Aegis 2.0), comprising 1,928 audio files generated from the test set prompts. This dataset enables multimodal AI safety research by providing spoken versions of adversarial and safety-critical prompts across 23 violation categories.
LANGUAGE: All prompts are in English. However, the audio files were… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-Content-Safety-Audio-Dataset.narrow-model-safety-eval
Narrow Model Safety Evaluation — Protein Dual-Use Risk Dataset
Summary: Annotations, results, and evaluation data for a proof-of-concept framework assessing dual-use risk in narrow scientific AI models. Two lines of work: (1) structure-level metrics — FSPE, FSI, and Physical Realizability Tier — on eight published protein toxins and mechanism-matched benign controls (ESM-2, ProteinMPNN); (2) mechanism generalization — a leave-one-mechanism-out panel measuring what an… See the full description on the dataset page: https://huggingface.co/datasets/jang1563/narrow-model-safety-eval.abuse-scanner-bot-datasetgender-secret-questions
Gender Secret Questions
Questions used to prompt-distil the gender secret model organisms.
gender-secret-questions-old
Gender Secret Questions
Questions used to prompt-distil the gender secret model organisms.
named-process-safety-incidents-extended-2026
Canonical landing page: https://www.smartqhse.com/datasets/named-process-safety-incidents-extended-2026
Named Process Safety and Industrial Disasters — Extended Reference 2026
Curated reference of 40 named historical process-safety, industrial, and major-fire disasters with dates, fatalities, casual factors, and regulatory consequences. Spans 1917–2024. Covers Bhopal, Piper Alpha, Texas City, Deepwater Horizon, Buncefield, Flixborough, Seveso, Phillips 66 Pasadena, Longford… See the full description on the dataset page: https://huggingface.co/datasets/SmartQHSE/named-process-safety-incidents-extended-2026.multilingual-elder-safety-msgs
multilingual-elder-safety-msgs
A hand-authored, multilingual elder fraud-recognition and safety coaching dataset. 467 curated scam/safe scenarios in Chinese and English, with platform-generated coaching responses localized across 5 languages: Chinese, English, Vietnamese, Khmer (Cambodian), and Lao. Expanded to 1,029 rows through Adaption Labs platform reasoning traces and multilingual adaptation.
Built for communities where filial piety, authority deference, and fear of… See the full description on the dataset page: https://huggingface.co/datasets/vanila434/multilingual-elder-safety-msgs.CSEI-SafetyBench
Chinese Explicit and Implicit Safety Benchmark
Dataset Description
The Chinese Explicit and Implicit Safety Benchmark is a collection of 1,000
Chinese prompts designed to evaluate safety risks in large language models.
It covers both directly expressed harmful requests and subtler risks that
depend on context, tone, implication, satire, or exaggeration.
The benchmark is intended for model safety evaluation, red-teaming, and
research on safety alignment in… See the full description on the dataset page: https://huggingface.co/datasets/BAAI/CSEI-SafetyBench.faa-aviation-safety-rollups
FAA wildlife strikes, laser incidents and drone sightings — analysis-ready rollups
Three United States FAA safety datasets, cleaned and rolled up into small tabular
files you can load without touching the source archives.
This is not a copy of the FAA's raw releases. Those are already public and
large. What is here is the part that does not exist upstream in this form:
stable slugs, consistent naming, and per-airport / per-species / per-state /
per-aircraft / per-year rollups… See the full description on the dataset page: https://huggingface.co/datasets/himaxym/faa-aviation-safety-rollups.conversionsmultilingual-safety
Multilingual Safety Instructions
A multilingual extension of the safety-only instruction–refusal pairs released with the Safety-Tuned LLaMAs project. The original 1,000 harmful-prompt / refusal-response pairs (English) were translated into 11 additional typologically diverse languages with google/gemini-2.0-flash-001. Each language is stored as a separate Hugging Face config.
Field
Description
prompt
Harmful user instruction (translated; en is the original).
output
Safe… See the full description on the dataset page: https://huggingface.co/datasets/iNLP-Lab/multilingual-safety.vn-provinces-safe-water-access-rate
Vietnam safe water access rate
Vietnam safe water access rate. Geographic labels are English (UN/GSO style ASCII romanization). Tables cover provinces, regions and national total where present. Province names follow ar_core.vn_geo (historical 63-province system).
Figures
Hero
Comparison
Color key
Files
provinces (441 rows)
data/provinces.csv
data/provinces.dta
data/provinces.xlsx
regions (42 rows)
data/regions.csv
data/regions.dta… See the full description on the dataset page: https://huggingface.co/datasets/letrinhan/vn-provinces-safe-water-access-rate.safer-instruct
Safer-Instruct: Aligning Language Models with Automated Preference Data
This repository contains the dataset for the paper titled "Safer-Instruct: Aligning Language Models with Automated Preference Data". Check out our project website here!
Abstract
Reinforcement learning from human feedback (RLHF) is a vital strategy for enhancing model capability in language models. However, annotating preference data for RLHF is a resource-intensive and creativity-demanding process… See the full description on the dataset page: https://huggingface.co/datasets/lime-nlp/safer-instruct.aart-ai-safety-datasetsafety-benchmark
Safety Classification Dataset
Dataset Summary
This dataset is designed for multi-label classification of text inputs, identifying whether they contain safety-related concerns. Each sample is labeled with one or more of the following categories:
Dangerous Content
Harassment
Sexually Explicit Information
Hate Speech
Safe
This Dataset contain 5000 samples.
Labeling Rules
If Safe = 0, at least one of the other labels (Dangerous Content, Harassment, Sexually… See the full description on the dataset page: https://huggingface.co/datasets/qualifire/safety-benchmark.Nemotron-SFT-Safety-v1-prompt-only
Nemotron-SFT-Safety-v1-prompt-only
Prompt-only extraction from nvidia/Nemotron-SFT-Safety-v1.
Files:
prompts.csv: one prompt extraction record per source row. Records include
prompt, separated system_prompt, and structured tools when the source row
defines available tools. Nested values are JSON-encoded inside CSV cells.
summary.md: source row counts, extracted row counts, count deltas, and failed prompt counts.
null_or_empty_rows.md: row indexes where prompt extraction… See the full description on the dataset page: https://huggingface.co/datasets/jamesdborin/Nemotron-SFT-Safety-v1-prompt-only.safeleak-rcd
SafeLeak-RCD: Residential Residual Current Decomposition Benchmark
This repository contains SafeLeak-RCD, the public benchmark bundle prepared for the manuscript:
Physics-Regularized Conditional Flow Matching for Branch-Conditioned Residual Current Decomposition in Electrical Safety Monitoring
Release Contents
benchmark/
The exact train/validation/test split used in the manuscript revision.
processed_entities/
The processed per-entity bundle used to construct the… See the full description on the dataset page: https://huggingface.co/datasets/haayan/safeleak-rcd.Nemotron-RL-Safety-v1-prompt-only
Nemotron-RL-Safety-v1-prompt-only
Prompt-only extraction from nvidia/Nemotron-RL-Safety-v1.
Files:
prompts.csv: one prompt extraction record per source row. Records include
prompt, separated system_prompt, and structured tools when the source row
defines available tools. Nested values are JSON-encoded inside CSV cells.
summary.md: source row counts, extracted row counts, count deltas, and failed prompt counts.
null_or_empty_rows.md: row indexes where prompt extraction produced… See the full description on the dataset page: https://huggingface.co/datasets/jamesdborin/Nemotron-RL-Safety-v1-prompt-only.SafeguardMTL
SafeguardMTL: Multilingual Dataset for AI Safety
SafeguardMTL is a curated, multilingual dataset designed to train Guardrail models (Safety Nodes) for Large Language Models (LLMs).
Unlike standard safety datasets, this dataset supports Multi-Task Learning (MTL) by providing three distinct layers of classification for every prompt: Context Category, User Intent, and Safety Risk. Additionally it includes a language label to filter or evaluate the performance on specific languages.
It… See the full description on the dataset page: https://huggingface.co/datasets/balidea-ai-lab/SafeguardMTL.SafeChem
Dataset Card for SafeChem
Dataset Summary
SafeChem is a regulatory-grounded benchmark dataset of 32,211
chemical substances for multi-label GHS (Globally Harmonized System)
hazard prediction and LLM safety reliability evaluation. Unlike prior
molecular benchmarks constructed by querying pharmaceutical databases,
SafeChem is seeded from a curated hazardous materials registry,
ensuring coverage of real-world industrial and safety-critical chemicals
including solvents… See the full description on the dataset page: https://huggingface.co/datasets/Anonymous-07/SafeChem.msha-mine-safety-violations-by-operator
MSHA Mine Safety Violations by Operator
Canonical version: The authoritative, most current version of this dataset lives at https://fastdol.com/datasets/msha-mine-safety-violations-by-operator. This Hugging Face copy is a mirror; refer to the canonical page for the latest data and documentation.
Mine safety enforcement records from the U.S. Mine Safety and Health Administration (MSHA), aggregated to the operator/contractor level. Every entity cited in MSHA's enforcement… See the full description on the dataset page: https://huggingface.co/datasets/FastDOLz/msha-mine-safety-violations-by-operator.safety-slice-auditcleaned_prompt_safety_datasetchemical-safety-boundary-recognition-v01Chemical Safety Boundary Recognition v01
What this dataset is
This dataset evaluates whether a system can recognize chemical danger before it escalates.
You give the model:
A reaction setup and scale
Hazards and limits
Live state and early warning signals
You ask it to choose one response.
This is not about being careful.
It is about seeing boundaries.
Why this matters
Many chemical incidents start with trends.
Temperature rising
Pressure oscillating
Gas evolving
Viscosity climbing
Equipment… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/chemical-safety-boundary-recognition-v01.vehicle-safety-profile
Vehicle Safety Profile — Under Review
New purchases paused.
Existing joined vehicle profile; timeline and derived flags are under review.
Verified coverage: Model years 2010–2023.
Records: 33,686 vehicle-year profiles.
This repository contains a 1,000-row public sample, not the full package. A sample does not establish complete historical coverage.
Limitations
New purchases are paused. The promised timeline table is not present.
Every complaint_trend is stable… See the full description on the dataset page: https://huggingface.co/datasets/claritystorm/vehicle-safety-profile.safe-storage-41a8da
safe-storage-41a8da
Synthetic weather test data: 34 rows in data.csv.
All values are randomly generated fictional examples, not real observations, products, or user activity. Intended only for CSV loading and pipeline tests; not suitable for scientific or business conclusions. Columns are sampled independently and do not model real-world correlations.
Fields
sample_id: random identifier for this generated sample.
row_id: sequential row number starting at 1.… See the full description on the dataset page: https://huggingface.co/datasets/cedarDawnJ/safe-storage-41a8da.safe-police-9935f1
safe-police-9935f1
Synthetic products test data: 36 rows in data.csv.
All values are randomly generated fictional examples, not real observations, products, or user activity. Intended only for CSV loading and pipeline tests; not suitable for scientific or business conclusions. Columns are sampled independently and do not model real-world correlations.
Fields
sample_id: random identifier for this generated sample.
row_id: sequential row number starting at 1.… See the full description on the dataset page: https://huggingface.co/datasets/zenithfield/safe-police-9935f1.grab-safe-driver-telematics-cleaned-datasetPKU-SafeRLHF-Prompts-Shift-answer-train-featuresAegis-Safety-DPO
Aegis: PolarAI's safety alignment dataset
Overview
Aegis-Safety-DPO is a high-density, (mostly) manually-curated preference dataset designed for Direct Preference Optimization (DPO) and Group Relative Policy Optimization (GRPO).
Unlike traditional safety datasets that train models to be "preachy," "evasive," or "apologetic", Aegis trains models that refuse to answer
Analyze the malicious request deeply using Chain-of-Thought (<think>).
Conclude objectively why the… See the full description on the dataset page: https://huggingface.co/datasets/PolarAI/Aegis-Safety-DPO.