SafeNLP: Trustworthy AI Through Rigorous Safety
Safety solutions and research for NLP, LLM, and AI
We explore safety solutions and research for NLP, LLM, and AI, designing approaches for both academic and industry levels with emphasis on security and ethical principles. Our work spans bias detection and mitigation, privacy-preserving techniques, adversarial robustness, content moderation, transparency, and comprehensive safety benchmarking for language models.