AI
Securing modern AI and using AI for security — from jailbreaking and safely fine-tuning large language models to ML/NLP methods that detect threats at scale.
KAIST · School of Electrical Engineering
Welcome to the NSS Lab @ KAIST
The Network and System Security (NSS) Laboratory, led by Prof. Seungwon Shin, reveals and understands emerging threats and designs new systems and algorithms that make computing secure.
Research Areas
The directions that drive our work.
Security
Revealing and defending against threats across networks, hardware, and blockchains — through deep system analysis, side-channel research, and threat forensics.
System
High-performance systems that are secure by construction — SmartNIC acceleration, confidential serverless, and secure container and in-network systems.
News
Recent publications and highlights.
- May 2026 PC²: Politically Controversial Content Generation via Jailbreaking Attacks on GPT-based Text-to-Image Models Accepted to CCS 2026
- February 2026 AccelFaaS: Accelerating FaaS via Pre-warmed Memory and Control Channel Offloading Accepted to TCC 2026
- February 2026 TDSnap: Enabling Secure Function-as-a-Service with Trusted Domain Snapshots Accepted to DAC 2026
- January 2026 SafeMoE: Safe Fine-Tuning for MoE LLMs by Aligning Harmful Input Routing Accepted to ICLR 2026
- December 2025 HybridMesh: A Hardware-software Hybrid Approach for Accelerating Service Mesh Ingress Accepted to NSDI 2026
- December 2025 RDNet: An RDMA-aware Container Network Interface for Cloud Environments Accepted to INFOCOM 2026
- November 2025 SECTRACER: A Framework for Uncovering the Root Causes of Network Intrusions via Security Provenance Accepted to Computers & Security 2025
- September 2025 MoEvil: Poisoning Expert to Compromise the Safety of Mixture-of-Experts LLMs Accepted to ACSAC 2025 — Distinguished Paper Award
- August 2025 Improbable Bigrams Expose Vulnerabilities of Incomplete Tokens in Byte-Level Tokenizers Accepted to EMNLP 2025
- June 2025 Refusal Is Not an Option: Unlearning Safety Alignment of Large Language Models Accepted to USENIX Security 2025
Find Us
NSS Laboratory
N1 Building, Room 820, KAIST, 291 Daehak-ro, Yuseong-gu, Daejeon, Korea
School of Electrical Engineering