$BKDR backdoor

Latest papers

1. RAGLeak: Supply-Chain Privacy Backdoors in Retrieval-Augmented Generation cs.CR · 19 Oct · 4 hits

   …on either prompt injection, which heavily depends on the generator's instruction-following capability, or backdoors that reproduce or paraphrase retrieved documents but struggle to balance extraction fidelity and…

2. Under the Skin: Backdoor Infection in Multimodal Large Language Models cs.CR · 16 Oct · 3 hits

   …performance in generating text conditioned on inputs from different modalities. In this work, we introduce backdoor infection, a cross-modal attack framework targeting MLLMs. Starting from a model already injected…

3. SemPurify: Semantics-Aware Data Purification for Black-box Backdoor Defense cs.CV · 16 Oct · 6 hits

   Backdoor attacks have emerged as a critical security threat in machine learning, where adversaries embed…

4. From Billion-Scale to Million-Scale Language Models: Operational Capabilities for Embedded Post-Exploitation cs.CR · 16 Oct · 2 hits

   …SLMs (20M-1.5B parameters) on two post-exploitation code-generation tasks: Directory Enumeration and Backdoor Generation. We compare Python (high-resource) and Lua (low-resource) across prompting strategies,…

5. Trojan Horse: Automated Repository Poisoning against Coding Agents cs.CR · 16 Oct · 1 hit

   …Extensive experiments show that TrojanForge achieves a stable poisoning success rate of over 60% with backdoor injection exceeding 90%. Comprehensive evaluation on TrojanBench across 4 harnesses and 12 models…

6. ShadowFPT: Backdooring Federated Prompt Tuning via Shadow Triggers cs.AI · 16 Oct · 3 hits

   …creates an overlooked security risk. Since the backbone is shared and frozen, a malicious client can induce backdoor behavior through the representation space while keeping its uploaded prompt updates close to benign…

7. SAFETYEDIT: Sparse Internal Targets with Prompt-Wide Control for LLM Jailbreak Backdoors cs.CL · 15 Oct · 2 hits

   Model-editing backdoors aim to induce harmful responses under a trigger while preserving ordinary behavior on untriggered…

8. Everyone's a little bit racist: a novel ML backdoor attack in black-box mode cs.CR · 15 Oct · 5 hits

   Backdoors are among the most impactful attacks against Machine Learning (ML), yet most existing methods…