📊 Dataset Collection Summary
Total Conversations
Total Turns
Dataset Types
Sample Data
📈 Full Dataset Statistics
| Dataset | Conversations | Turns | Avg Turns/Conv | Focus |
|---|---|---|---|---|
| Original Multi-turn | 594+ | 4,642+ | 7.8 | Baseline organic conversations |
| └ Psychology | 158+ | 1,583+ | 10.0 | Psychology harm category |
| └ Illicit | 436+ | 3,059+ | 7.0 | Illicit harm category |
| Bio-transformed V1 | 1,309+ | 6,784+ | 5.2 | Direct bio-safety attacks |
| Bio-transformed V2 | 1,308+ | 8,127+ | 6.2 | Adaptive bio-safety attacks |
| Keyword-transformed | 7,110+ | 53,705+ | 7.6 | Cross-domain harm transfer |
🔗 Access Datasets on Hugging Face
Original Multi-turn Conversations
Psychology + Illicit baseline conversations
Sample: 50 conversations, 390 turns
Bio-transformed Synthetic V1
Direct bio-topic transformation methodology
Sample: 50 conversations, 449 turns
Bio-transformed Synthetic V2
Adaptive bio-topic transformation methodology
Sample: 50 conversations, 459 turns
Keyword-transformed Synthetic
Cross-domain keyword substitution methodology
Sample: 50 conversations, 659 turns
🧪 Research Applications
Safety Evaluation
- Benchmark model safety
- Measure robustness
- Evaluate mechanisms
Red Teaming
- Discover adversarial patterns
- Test safety guardrails
- Identify blind spots
Model Training
- Fine-tune safety classifiers
- Train attack detectors
- Develop harm detection
Safety Research
- Study harm transfer
- Analyze attack patterns
- Understand dynamics
⚠️ Ethical Considerations
⚠️ IMPORTANT
These datasets contain successful adversarial attacks and harmful content.
✅ Intended Use
- Defensive security research
- AI safety evaluation and improvement
- Academic research on adversarial robustness
- Training safety and moderation systems
❌ Prohibited Use
- Creating offensive content
- Developing attack tools for malicious purposes
- Bypassing safety systems for harm
- Any use that violates laws or ethical guidelines
🎯 Data Selection Process
All datasets are derived from high-quality, validated conversations with strict quality filters including NeurIPS evaluation protocols.
Base Criteria
- Text-based objectives (no code execution templates)
- NeurIPS evaluation metadata present
- Verdict:
success(harmful requests successfully fulfilled) - Multi-turn conversations with prompt-response pairs
Psychology-Specific Criteria
- Organic conversations (
organicity = true) - No disclaimer in responses
- Successfully elicited harmful psychology-related content
Illicit-Specific Criteria
- Contains specific instruction details
- Practically executable (not abstract)
- Successfully elicited harmful illicit-related content
📄 License
All datasets are released under CC-BY-NC-4.0 (Creative Commons Attribution-NonCommercial 4.0 International).
- ✅ Use for research and evaluation
- ✅ Modify and build upon the data
- ✅ Share with attribution
- ❌ Commercial use without separate licensing
💼 Full Dataset Access
The sample datasets provide representative examples. Full datasets contain thousands of additional conversations with expanded harm categories and regular updates.
Please contact us at info@gojuly.ai to purchase any or all of full datasets.
Include your research objectives, institutional affiliation, and intended use in your inquiry.