ISACL: Internal State Analyzer for Copyrighted Training Data Leakage

Guangwei Zhang, Qisheng Su, Jiateng Liu +4

Large Language Models (LLMs) have revolutionized Natural Language Processing (NLP) but pose risks of inadvertently exposing copyrighted or proprietary data, especially when such da…