The Alignment Curse: Modality Alignment Supercharges Audio Attacks via Text Transfer
cs.LG, cs.AI, cs.SD
Submitted: 2026-01-30
Updated: 2026-11-13
Terminology
Sources
- GPT-4 Technical Report
- VioLA: Unified Codec Language Models for Speech Recognition, Synthesis, and Translation
- Qwen2-Audio Technical Report
- GPT-4o System Card
- Qwen3-Omni Technical Report
- LongCat-Flash-Omni Technical Report
- VoiceBench: Benchmarking LLM-Based Voice Assistants
- Universal and Transferable Adversarial Attacks on Aligned Language Models
- Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities
- FunAudioLLM: Voice Understanding and Generation Foundation Models for Natural Interaction Between Humans and LLMs
- Qwen-Audio: Advancing Universal Audio Understanding via Unified Large-Scale Audio-Language Models
- Qwen2.5-Omni Technical Report
- InteractiveOmni: A Unified Omni-modal Model for Audio-Visual Multi-turn Dialogue
- Jailbreak Attacks and Defenses Against Large Language Models: A Survey
- D squared-Monitor: Dynamic Safety Monitoring for Diffusion LLMs via Hesitation Signals
- FORCE: Transferable Visual Jailbreaking Attacks via Feature Over-Reliance CorrEction
- Voice Jailbreak Attacks Against GPT-4o
- Speech-Audio Compositional Attacks on Multimodal LLMs and Their Mitigation with SALMONN-Guard
- Large Language Models Encode Semantics and Alignment in Linearly Separable Representations
- A Comprehensive Survey in LLM(-Agent) Full Stack Safety: Data, Training and Deployment
Related papers
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis
- (Mis)generalization of Helpful-Only Fine-tuning