Michael Backes

Benchmarking Empirical Privacy Protection for Adaptations of Large Language Models

Excessive Reasoning Attack on Reasoning LLMs

Benchmark of Benchmarks: Unpacking Influence and Code Repository Quality in LLM Safety Benchmarks

TrustGen: A Platform of Dynamic Benchmarking on the Trustworthiness of Generative Foundation Models

SafeReview: Building a Robust Deep Review Assistant Against Prompt Injection

Sparse Models, Sparse Safety: Unsafe Routes in Mixture-of-Experts LLMs

PRIVDISTIL: A Unified Framework for Accurate and Differentially Private Model Compression

Boosting Safety Alignment in LLMs with Response Shortcuts

IAAgent: Autonomous Inference Attacks Against ML Services With LLM-Based Agents

SOS! Soft Prompt Attack Against Open-Source Large Language Models

Finding and Reactivating Post-Trained LLMs' Hidden Safety Mechanisms

Adjacent Words, Divergent Intents: Jailbreaking Large Language Models via Task Concurrency

SaLoRA: Safety-Alignment Preserved Low-Rank Adaptation

Provably Cost-Sensitive Adversarial Defense via Randomized Smoothing

Efficient and Privacy-Preserving Soft Prompt Transfer for LLMs

Captured by Captions: On Memorization and its Mitigation in CLIP Models

POST: A Framework for Privacy of Soft-prompt Transfer

ACE: Attack Combo Enhancement Against Machine Learning Models