Prompt Injection datasets gathered from Gandalf (https://gandalf.lakera.ai/) and other prompt injection CTFs we've run for specific events.
Lakera
company
Verified
AI & ML interests
AI Safety, Computer Vision, NLP, Responsible AI, AI Fairness, Model validation
Collections
2
A collection of datasets and papers discussed during our "Lessons Learned from Crowdsourced LLM Threat Intelligence" webinar.
-
Lakera/gandalf_ignore_instructions
Viewer • Updated • 1k • 277 • 26 -
Lakera/gandalf_summarization
Viewer • Updated • 140 • 83 • 3 -
Ignore This Title and HackAPrompt: Exposing Systemic Vulnerabilities of LLMs through a Global Scale Prompt Hacking Competition
Paper • 2311.16119 • Published • 2 -
hackaprompt/hackaprompt-dataset
Viewer • Updated • 602k • 497 • 42
models
5
Lakera/autotrain-cancer-lakera-50807121085
Image Classification
•
Updated
•
6
Lakera/autotrain-cancer-lakera-50807121082
Image Classification
•
Updated
Lakera/autotrain-cancer-lakera-50807121084
Image Classification
•
Updated
•
5
Lakera/autotrain-cancer-lakera-50807121083
Image Classification
•
Updated
•
5
Lakera/autotrain-cancer-lakera-50807121081
Image Classification
•
Updated
•
6
datasets
5
Lakera/gandalf-trial-data-benign
Viewer
•
Updated
•
18k
•
17
Lakera/mosscap_prompt_injection
Viewer
•
Updated
•
279k
•
223
•
8
Lakera/gandalf_ignore_instructions
Viewer
•
Updated
•
1k
•
277
•
26
Lakera/gandalf_summarization
Viewer
•
Updated
•
140
•
83
•
3
Lakera/autotrain-data-cancer-lakera
Preview
•
Updated
•
34
•
3