LegalGuardian masks PII in legal prompts using NER and a local LLM, achieving 93% to 97% F1 on a small synthetic immigration dataset, but the evaluation is undermined by same-model circularity.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
LegalGuardian: A Privacy-Preserving Framework for Secure Integration of Large Language Models in Legal Practice
LegalGuardian masks PII in legal prompts using NER and a local LLM, achieving 93% to 97% F1 on a small synthetic immigration dataset, but the evaluation is undermined by same-model circularity.