Image de couverture de livre

CaMeLs Can Use Computers Too: System-level Security for Computer Use Agents

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Foerster, Hanna, Blanchard, Tom, Nikolić, Kristina, Shumailov, Ilia, Zhang, Cheng, Mullins, Robert, Papernot, Nicolas, Tramèr, Florian, Zhao, Yiren
Format:	Preprint
Publié:	2026
Sujets:	Artificial Intelligence
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

Architectural Neural Backdoors from First Principles
par: Langford, Harry, et autres
Publié: (2024)

Hardware and Software Platform Inference
par: Zhang, Cheng, et autres
Publié: (2024)

Quantamination: Dynamic Quantization Leaks Your Data Across the Batch
par: Foerster, Hanna, et autres
Publié: (2026)

Beyond Slow Signs in High-fidelity Model Extraction
par: Foerster, Hanna, et autres
Publié: (2024)

Reasoning Introduces New Poisoning Attacks Yet Makes Them More Complicated
par: Foerster, Hanna, et autres
Publié: (2025)

Fairness Feedback Loops: Training on Synthetic Data Amplifies Bias
par: Wyllie, Sierra, et autres
Publié: (2024)

The Curse of Recursion: Training on Generated Data Makes Models Forget
par: Shumailov, Ilia, et autres
Publié: (2023)

Modal Aphasia: Can Unified Multimodal Models Describe Images From Memory?
par: Aerni, Michael, et autres
Publié: (2025)

ImpNet: Imperceptible and blackbox-undetectable backdoors in compiled neural networks
par: Clifford, Eleanor, et autres
Publié: (2022)

Operationalizing CaMeL: Strengthening LLM Defenses for Enterprise Deployment
par: Tallam, Krti, et autres
Publié: (2025)

Beyond Labeling Oracles: What does it mean to steal ML models?
par: Shafran, Avital, et autres
Publié: (2023)

Beyond Laplace and Gaussian: Exploring the Generalized Gaussian Mechanism for Private Machine Learning
par: Rinberg, Roy, et autres
Publié: (2025)

Breach By A Thousand Leaks: Unsafe Information Leakage in `Safe' AI Responses
par: Glukhov, David, et autres
Publié: (2024)

Inexact Unlearning Needs More Careful Evaluations to Avoid a False Sense of Privacy
par: Hayes, Jamie, et autres
Publié: (2024)

Gradients Look Alike: Sensitivity is Often Overestimated in DP-SGD
par: Thudi, Anvith, et autres
Publié: (2023)

When Vision Fails: Text Attacks Against ViT and OCR
par: Boucher, Nicholas, et autres
Publié: (2023)

RealMath: A Continuous Benchmark for Evaluating Language Models on Research-Level Mathematics
par: Zhang, Jie, et autres
Publié: (2025)

The Jailbreak Tax: How Useful are Your Jailbreak Outputs?
par: Nikolić, Kristina, et autres
Publié: (2025)

Watermarking Needs Input Repetition Masking
par: Khachaturov, David, et autres
Publié: (2025)

Locking Machine Learning Models into Hardware
par: Clifford, Eleanor, et autres
Publié: (2024)

Revisiting Block-based Quantisation: What is Important for Sub-8-bit LLM Inference?
par: Zhang, Cheng, et autres
Publié: (2023)

A Systematization of Security Vulnerabilities in Computer Use Agents
par: Jones, Daniel, et autres
Publié: (2025)

Scaling Agents for Computer Use
par: Gonzalez-Pumariega, Gonzalo, et autres
Publié: (2025)

On the Reliability of Computer Use Agents
par: Gonzalez-Pumariega, Gonzalo, et autres
Publié: (2026)

Multi-Agent Computer Use
par: Koh, Jing Yu, et autres
Publié: (2026)

Programming with Pixels: Can Computer-Use Agents do Software Engineering?
par: Aggarwal, Pranjal, et autres
Publié: (2025)

Thought-Transfer: Indirect Targeted Poisoning Attacks on Chain-of-Thought Reasoning Models
par: Chaudhari, Harsh, et autres
Publié: (2026)

ceLLMate: Sandboxing Browser AI Agents
par: Meng, Luoxi, et autres
Publié: (2025)

Grounding Computer Use Agents on Human Demonstrations
par: Feizi, Aarash, et autres
Publié: (2025)

Secure and Efficient Access Control for Computer-Use Agents via Context Space
par: Gong, Haochen, et autres
Publié: (2025)

Efficient Agent Training for Computer Use
par: He, Yanheng, et autres
Publié: (2025)

Evaluating the Robustness of the "Ensemble Everything Everywhere" Defense
par: Zhang, Jie, et autres
Publié: (2024)

Architectural Backdoors for Within-Batch Data Stealing and Model Inference Manipulation
par: Küchler, Nicolas, et autres
Publié: (2025)

AgentSentinel: An End-to-End and Real-Time Security Defense Framework for Computer-Use Agents
par: Hu, Haitao, et autres
Publié: (2025)

Heterogeneous Computing: The Key to Powering the Future of AI Agent Inference
par: Zhao, Yiren, et autres
Publié: (2026)

Mapping the Design Space of User Experience for Computer Use Agents
par: Cheng, Ruijia, et autres
Publié: (2026)

Direct-Scoring NLG Evaluators Can Use Pairwise Comparisons Too
par: Lawrence, Logan, et autres
Publié: (2025)

OpenComputer: Verifiable Software Worlds for Computer-Use Agents
par: Wei, Jinbiao, et autres
Publié: (2026)

Small Colleges Can Use Computers in Big Way.
par: Squires, Burton E., Jr.
Publié: (1968)

UnUnlearning: Unlearning is not sufficient for content regulation in advanced generative AI
par: Shumailov, Ilia, et autres
Publié: (2024)