Saved in:
Bibliographic Details
Main Authors: Mohammad, Noor Islam S., Bayazıt, Uluğ
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.16606
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917417710518272
author Mohammad, Noor Islam S.
Bayazıt, Uluğ
author_facet Mohammad, Noor Islam S.
Bayazıt, Uluğ
contents Large language models (LLMs) are increasingly deployed in high-stakes domains, yet a unified treatment of their overlapping safety challenges remains lacking. We present SafeLM, a framework that jointly addresses four pillars of LLM safety: privacy, security, misinformation, and adversarial robustness. SafeLM combines federated training with gradient smartification and Paillier encryption for privacy, integrates defenses against training and inference-time attacks, employs contrastive grounding with calibrated decoding to reduce hallucinations, and introduces alignment-aware binarized aggregation to enhance robustness while maintaining bounded reconstruction quality. Across benchmarks on factuality, toxicity, and membership inference, SafeLM achieves 98.0% harmful content detection accuracy, reduces communication by 96.9%, and lowers gradient inversion PSNR from 31.7 dB to 15.1 dB. Ablations show that each component contributes independently, whereas their integration yields a strong privacy utility efficiency trade-off for deploying trustworthy LLMs.
format Preprint
id arxiv_https___arxiv_org_abs_2604_16606
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle SafeLM: Unified Privacy-Aware Optimization for Trustworthy Federated Large Language Models
Mohammad, Noor Islam S.
Bayazıt, Uluğ
Cryptography and Security
Machine Learning
Large language models (LLMs) are increasingly deployed in high-stakes domains, yet a unified treatment of their overlapping safety challenges remains lacking. We present SafeLM, a framework that jointly addresses four pillars of LLM safety: privacy, security, misinformation, and adversarial robustness. SafeLM combines federated training with gradient smartification and Paillier encryption for privacy, integrates defenses against training and inference-time attacks, employs contrastive grounding with calibrated decoding to reduce hallucinations, and introduces alignment-aware binarized aggregation to enhance robustness while maintaining bounded reconstruction quality. Across benchmarks on factuality, toxicity, and membership inference, SafeLM achieves 98.0% harmful content detection accuracy, reduces communication by 96.9%, and lowers gradient inversion PSNR from 31.7 dB to 15.1 dB. Ablations show that each component contributes independently, whereas their integration yields a strong privacy utility efficiency trade-off for deploying trustworthy LLMs.
title SafeLM: Unified Privacy-Aware Optimization for Trustworthy Federated Large Language Models
topic Cryptography and Security
Machine Learning
url https://arxiv.org/abs/2604.16606