Alina Oprea

Localize-and-Detect: Auditing Task-Level Poisoning in Instruction-Tuned Models

The Privacy Fallacy of Crowdsourced Fine-Tuning: Extracting Proprietary Data via Topic-Based Poisoning

Beyond Predictable Paths: Redefining AI Security Incident Reporting for Agents

BlueSTAR: Tiered Agentic Architecture for Autonomous Cyber Defense

Targeting World Models to Compromise Robot Learning Pipelines

A Bayesian Approach to Membership Inference for Statistical Release

PoisonForge: Task-Level Targeted Poisoning Benchmark for Instruction-Tuned LLMs

Who Owns This Agent? Tracing AI Agents Back to Their Owners

Attacks and Mitigations for Distributed Governance of Agentic AI under Byzantine Adversaries

Reconstruction of Personally Identifiable Information from Proprietary Data in Supervised Fine-Tuned Models

MAGIQ: A Post-Quantum Multi-Agentic AI Governance System with Provable Security

Toward a Principled Framework for Agent Safety Measurement

Retrieval-Augmented LLMs for Security Incident Analysis

MUZZLE: Adaptive Agentic Red-Teaming of Web Agents Against Indirect Prompt Injection Attacks

Beware Untrusted Simulators -- Reward-Free Backdoor Attacks in Reinforcement Learning

Syntax- and Compilation-Preserving Evasion of LLM Vulnerability Detectors

Thought-Transfer: Indirect Targeted Poisoning Attacks on Chain-of-Thought Reasoning Models

Identifying Models Behind Text-to-Image Leaderboards