Michael Backes

GeoCloak: An Agentic Framework for Semantics-Preserving Geolocation Privacy via Peripheral Clue Manipulation

MemeTrace: A Multimodal LLM-Based Agent for Safety-Aware Meme Provenance Analysis

No Free Efficiency: Revisiting the Trade-off Between Training Efficiency and Model Vulnerability

AgentProv: Auditing Agentic LLM API Providers via Tool-use Policy Probes

Benchmarking Empirical Privacy Protection for Adaptations of Large Language Models

BadBone: Backdoor Attacks Against Backbone Models in Visual Prompt Learning

Trust Me, Import This: Dependency Steering Attacks via Malicious Agent Skills

Pop Quiz Attack: Black-box Membership Inference Attacks Against Large Language Models

SafeReview: Defending LLM-based Review Systems Against Adversarial Hidden Prompts

A Systematic Study of Training-Free Methods for Trustworthy Large Language Models

Pruning Unsafe Tickets: A Resource-Efficient Framework for Safer and More Robust LLMs

HarmfulSkillBench: How Do Harmful Skills Weaponize Your Agents?

TrustLDM: Benchmarking Trustworthiness in Language Diffusion Models

The Art of (Mis)alignment: How Fine-Tuning Methods Effectively Misalign and Realign LLMs in Post-Training

When Understanding Becomes a Risk: Authenticity and Safety Risks in the Emerging Image Generation Paradigm

Understanding LLM Behavior When Encountering User-Supplied Harmful Content in Harmless Tasks

Finding and Reactivating Post-Trained LLMs’ Hidden Safety Mechanisms

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

Real Money, Fake Models: Deceptive Model Claims in Shadow APIs

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

"Humans welcome to observe": A First Look at the Agent Social Network Moltbook

Jailbreaking Attacks vs. Content Safety Filters: How Far Are We in the LLM Safety Arms Race?

GEO-Detective: Unveiling Location Privacy Risks in Images with LLM Agents

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