Library/tensor-trust
Library/tensor-trustForked

HumanCompatibleAI/tensor-trust

tensor-trust

A prompt injection game to collect data for robust ML research

Builder

HumanCompatibleAI

HumanCompatibleAI

HumanCompatibleAI • individual

Stars

69

Using upstream star count

Forks

8

Using upstream fork count

Open Issues

0

Activity Score

0/100

0 commits in 30d

Created

Jun 5, 2023

Project creation date

README Summary

Tensor Trust is a prompt injection game designed to collect adversarial data for machine learning research focused on AI safety and robustness. Players attempt to extract secret information from AI systems through creative prompting techniques, while defenders try to protect against these attacks. The collected data helps researchers understand and improve AI systems' resistance to prompt injection vulnerabilities.

AI Dev Skills

Unmapped

Prompt EngineeringAdversarial Machine LearningAI Safety ResearchPrompt Injection DetectionLanguage Model SecurityHuman-AI InteractionCrowdsourced Data CollectionAdversarial Prompt Generation

Tags

Prompt EngineeringAdversarial Machine LearningAI Safety ResearchPrompt Injection DetectionLanguage Model SecurityHuman-AI InteractionCrowdsourced Data CollectionAdversarial Prompt GenerationSelf-hostedResearchAI SafetyLanguage Model Robustness TestingCloud APIPrompt Injection Defense ResearchAI Safety Training Data GenerationAdversarial AITextHuman-AI CollaborationAdversarial Prompt Data CollectionEducationCrowdsourced Security ResearchPython

Taxonomy

Recent Activity

Updated 1 years ago

7 Days

0

30 Days

0

90 Days

0

Quality

research
Quality
medium
Maturity
research

Categories

Dev Tools & AutomationPrimaryLearning ResourcesSafety & AlignmentSearch & KnowledgeOther AI / MLAI AgentsModel Training

PM Skills

Developer Platform

Languages

Python100.0%

Timeline

Project created
Jun 5, 2023
Forked
Mar 21, 2026
Your last push
1 years ago
Upstream last push
1 years ago
Tracked since
Jan 27, 2025

Similar Repos

pgvector cosine similarity · $0

Loading…