TY  - RPRT
TI  - Provably Protecting Fine-Tuned LLMs from Training Data Extraction while Preserving Utility
AU  - Tom Segal
AU  - Asaf Shabtai
AU  - Yuval Elovici
PY  - 2026
UR  - https://arxiv.org/abs/2602.00688
ID  - 2602.00688
ER  -