LLM semantic authentication accepts 99.5% of legitimate non-exact answers at 0.1% false-accept rate while RAG fraud detection lowers false positives from 17.2% to 3.5%.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CR 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
LLM-Assisted Authentication and Fraud Detection
LLM semantic authentication accepts 99.5% of legitimate non-exact answers at 0.1% false-accept rate while RAG fraud detection lowers false positives from 17.2% to 3.5%.