Data Security in Large Language Models: Risks, Defense, and Directions
cs.CR, cs.AI
Submitted: 2025-08-04
Updated: 2026-09-14
Comments: Published in Journal of King Saud University Computer and Information Sciences; substantially revised relative to v1
Journal ref: J. King Saud Univ. Comput. Inf. Sci. 38, 768 (2026)
DOI: 10.1007/s44443-026-01200-9
Code: https://github.com/neulab/RIPPLe
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Language (Technology) is Power: A Critical Survey of "Bias" in NLP
- On the Opportunities and Risks of Foundation Models
- Discovering Latent Knowledge in Language Models Without Supervision
- AutoHall: Automated Factuality Hallucination Dataset Generation for Large Language Models
- Killing One Bird with Two Stones: Model Extraction and Attribute Inference Attacks against BERT-based APIs
- Integrating Stock Features and Global Information via Large Language Models for Enhanced Stock Return Prediction
- Documenting Large Webtext Corpora: A Case Study on the Colossal Clean Crawled Corpus
- Differentially Private Continual Learning
- What Doesn't Kill You Makes You Robust(er): How to Adversarially Train against Data Poisoning
- On Large Language Models' Hallucination with Regard to Known Facts
- Turning Generative Models Degenerate: The Power of Data Poisoning Attacks
- Weight Poisoning Attacks on Pre-trained Models
- Decoding Knowledge Attribution in Mixture-of-Experts: A Framework of Basic-Refinement Collaboration and Efficiency Analysis
- BadEdit: Backdooring large language models by model editing
- Prompt Injection attack against LLM-integrated Applications
- Backdoor attacks and defenses in feature-partitioned collaborative learning
- It's All in the Name: Mitigating Gender Bias with Name-Based Counterfactual Data Substitution
- Inverse Scaling: When Bigger Isn't Better
- On the Risk of Misinformation Pollution with Large Language Models
- Detection of Adversarial Training Examples in Poisoning Attacks through Anomaly Detection
Related papers
- SoK: AI-Augmented Binary Reversing
- Relaxed Sender Anonymity for CBDC Interbank Settlement: A Zero-Knowledge Approach on Permissioned EVM
- Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages
- Efficient Fuzzy PSI under One-Sided Assumptions
- Sealing the Audit-Runtime Gap for LLM Skills
- Token Composition: A Graph Based on EVM Logs