NeurIPS 2026 reviews are up: reviews, scores, rebuttals and meta-reviews for 7,877 papers. See the highest rated
acceptodds: the papers of ICLR 2027
341 results for “keyword:"attacks"” in ICLR 2027. 100 withdrawn or desk rejected listed apart. clearsort:relevanceratingnewestmost read
[PDF]Chat Templates Are a Hidden Instruction Channel: Inference-Time Backdoors in LLM-Based Systems▲4002.67
[PDF]ARMOR: Agentic Red-Teaming with Memory-Optimized Retrieval for Autonomous Jailbreaking of Frontier Language Models▼1601.50
[PDF]ScaleTrojan: Uncovering Backdoor Vulnerability in Visual Autoregressive Text-to-Image Generation▲4002.67
[PDF]ADAN: Adaptive Direction-Aware Noise for Defending Iterative Gradient Inversion Attacks▼1.2k1.67
[PDF]SimAttack: A Simple and Effective Transferable Untargeted Adversarial Attack on Large Vision-Language Models▲6.1k2.00
[PDF]Cross-Encoder Fairness Degradation in Text-Attributed Graphs via Adversarial Node Injection▲9.2k2.67
[PDF]Restoring Backdoor Detection under Adaptive Attacks in Multimodal Large Language Models▼1.3k1.75
[PDF]Prune Less, Pay More: Imperceptible Budget Inflation Attacks on Token Pruning in Vision-Language Models▼1.2k1.67
[PDF]CamouPurifier: Zero-shot and Camouflage-Agnostic Defense Against Adversarial Camouflage Attacks on Monocular Depth Estimation▲5103.00
[PDF]AutoSkillDrain: Resource Amplification Attacks on LLM Agents via Malicious Skill Optimization▲3.0k2.33
[PDF]TARGETED FORGETTING: A FRAMEWORK FOR CLIENT- CLASS CENTRIC KNOWLEDGE REMOVAL IN FEDERATED LEARNING▼4.3k2.00
[PDF]Morpheus: Modality-Agnostic Adversarial Attack from Leaked Models in Federated Learning▲5.4k3.00
[PDF]Never Sampled, Always Accepted: Commitment Injection Through Top- Verification in Speculative Decoding▲4002.67
[PDF]Twenty-Five Samples to Blind a Guardrail: Stealthy Benign-Side Poisoning of Self-Adapting Safety Filters▲6.1k2.00
[PDF]EchoGhost: A Frequency-Structural Backdoor Attack and Spectral-Counterfactual Defense for Graph Neural Networks▼1351.33
[PDF]Beyond Image-Wide Color Shifts: Scene-Compatible Semantic-Region Triggers for Backdoor Attacks▼1351.33
Withdrawn or desk rejected (100)
[PDF]Attested but Unaccounted: DP-SGD Privacy Accounting as a Precondition of Every StepWithdrawn1.67
[PDF]TypoShift: Open-Set Retrieval Hijacking and Attention Pathway Surgery for Text-to-Image Person Re-IdentificationWithdrawn2.00
[PDF]HAFT: Adversarial Robustness Benchmarking for Hindi Fact-Checking with Adaptive Attack SelectionDesk rejected
[PDF]It Takes Two to Tango: Backdoor Attack for Multimodal Diffusion Models via Cross-modal Collaborative TriggersWithdrawn1.67
[PDF]WMAttack: Automated Attack Search for Adversarial Evaluation of World-Model AgentsWithdrawn1.67