Maximilian Baader

Researcher@Snyk·瑞士·OpenReview

研究方向

Agents · large language models · code security · federated learning · privacy attacks · adversarial robustness · certified robustness · neural network verification

Maximilian Baader

ToolFuzz: Automated Agent Tool Testing

Adaptive Generation of Bias-Eliciting Questions for LLMs

CommandSans: Securing AI Agents with Surgical Precision Prompt Sanitization

AutoBaxBench: Bootstrapping Code Security Benchmarking

Polyrating: A Cost-Effective and Bias-Aware Rating System for LLM Evaluation

Ward: Provable RAG Dataset Inference via LLM Watermarks

A Unified Approach to Routing and Cascading for LLMs

Certified Robustness to Data Poisoning in Gradient-Based Training

GRAIN: Exact Graph Reconstruction from Gradients

A Unified Approach to Routing and Cascading for LLMs

BaxBench: Can LLMs Generate Correct and Secure Backends?

Evading Data Contamination Detection for Language Models is (too) Easy

Gaussian Loss Smoothing Enables Certified Training with Tight Convex Relaxations