CVPR 2026 accepted papers on Adversarial robustness and security
76 of the 4,042 papers accepted at CVPR 2026 (1.9%) are on Adversarial robustness and security, compared with 2.2% at CVPR 2025. They fall into 4 topics.
See CVPR 2026 in the interactive map
Working on something in this area? Describe your idea in scime atlas to see the closest accepted papers by meaning and how crowded the spot has become.
Topics and papers
Attacks on text and graphs · 40 papers
- All Vehicles Can Lie: Efficient Adversarial Defense in Fully Untrusted-Vehicle Collaborative Perception via Pseudo-Random Bayesian Inference (1 citation)
- A Provable Energy-Guided Test-Time Defense Boosting Adversarial Robustness of Large Vision-Language Models
- Adversarial Style Optimization: Enhancing VLM Jailbreaks by GRPO-based Stylistic Triggers Optimization
- AGFT: Alignment-Guided Fine-Tuning for Zero-Shot Adversarial Robustness of Vision-Language Models
- ARGUS: Defending Against Multimodal Indirect Prompt Injection via Steering Instruction-Following Behavior
- Attack for Defense: Adversarial Agents for Point Prompt Optimization Empowering Segment Anything Model
- Do Vision-Language Models Leak What They Learn? Adaptive Token-Weighted Model Inversion Attacks
- FlowHijack: A Dynamics-Aware Backdoor Attack on Flow-Matching Vision-Language-Action Models
- Hierarchically Robust Zero-shot Vision-language Models
- LaSM: Layer-wise Scaling Mechanism for Defending Pop-up Attack on GUI Agents
- Learning Mutual View Information Graph for Adaptive Adversarial Collaborative Perception
- Out of Sight, Out of Track: Adversarial Attacks on Propagation-based Multi-Object Trackers via Query State Manipulation
and 28 more in the interactive map.
Black-box and transfer attacks · 16 papers
- Black-box Membership Inference Attacks on the Pre-training Data of Image-generation Models
- DASH: A Meta-Attack Framework for Synthesizing Effective and Stealthy Adversarial Examples
- FeatureFool: Zero-Query Fooling of Video Models via Feature Map
- Generative Adversarial Perturbations with Cross-paradigm Transferability on Localized Crowd Counting
- Multi-Paradigm Collaborative Adversarial Attack Against Multi-Modal Large Language Models
- PA-Attack: Guiding Gray-Box Attacks on LVLM Vision Encoders with Prototypes and Attention
- RaPA: Enhancing Transferable Targeted Attacks via Random Parameter Pruning
- SEBA: Sample-Efficient Black-Box Attacks on Visual Reinforcement Learning
- Shedding Light on VLN Robustness: A Black-box Framework for Indoor Lighting-based Adversarial Attack
- What Your Features Reveal: Data-Efficient Black-Box Feature Inversion Attack for Split DNNs
- AdvFM: Lookahead Flow-Matching Velocity-Field Attacks for Imperceptible and Transferable Adversarial Examples
- CamPI: Physical Adversarial Examples through Camera Power Signal Injection
and 4 more in the interactive map.
Certified and empirical robustness · 14 papers
- A Combination of Noise and Bilateral Filters Achieve Supralinear and Scalable Adversarial Robustness in CNNs
- A Unified Perspective on Adversarial Membership Manipulation in Vision Models
- Decoupling Defense Strategies for Robust Image Watermarking
- IrisFP: Adversarial-Example-based Model Fingerprinting with Enhanced Uniqueness and Robustness
- Lipschitz Optimization for Formal Verification of Homographies
- Mitigating Error Amplification in Fast Adversarial Training
- Taming the Long Tail: Rebalancing Adversarial Training via Adaptive Perturbation
- The Power of Decaying Steps: Enhancing Attack Stability and Transferability for Sign-based Optimizers
- Towards Reliable Evaluation of Adversarial Robustness for Spiking Neural Networks
- ReMoE: Region-Mixture Experts for Adversarially-Robust Vision Transformers
- Robustness Under Data Scarcity: Few-Shot Continual Adversarial Training for Evolving Threats
- Towards Robust Vision Transformers: Path Dependency Analysis and a Simple Two-Stage Adversarial Training
and 2 more in the interactive map.
Backdoor and poisoning · 6 papers
- BlackMirror: Black-Box Backdoor Detection for Text-to-Image Models via Instruction-Response Deviation
- Exposing Functional Fusion: A New Class of Strategic Backdoor in Dynamic Prompt Architectures
- Towards Stealthy and Effective Backdoor Attacks on Lane Detection: A Naturalistic Data Poisoning Approach
- Logit-Margin Repulsion for Backdoor Defense
- Phantom: Physical Object Interactions as Dynamic Triggers for NMS-Exploited Backdoors
- Unleashing Stealthy Backdoor Pandemic by Infecting a Single Diffusion Model