# CASA: Classification Augmented with Safety Attention for Robust Multimodal Alignment

arXiv: https://arxiv.org/abs/2604.00310 · HTML: https://mlsanity.com/paper/2604.00310/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Safety alignment and jailbreak attacks, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 19 of its 37 references that are in the map.
- First posted on arXiv 31 Mar 2026.

## Placed in

- LLM alignment / RLHF / preference optimisation / [Safety alignment and jailbreak attacks](https://mlsanity.com/alignment/safety-alignment-and-jailbreak-attacks/)
- LLM safety and security / [Vision-language and text-to-image safety](https://mlsanity.com/safety/vision-language-and-text-to-image-safety/)

## Builds on (most cited first, 19 of 19)

- [Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!](https://mlsanity.com/paper/2310.03693/) (2023)
- [Aligning AI With Shared Human Values](https://mlsanity.com/paper/2008.02275/) (2020)
- [Parameter-Efficient Fine-Tuning for Large Models: A Comprehensive Survey](https://mlsanity.com/paper/2403.14608/) (2024)
- [Jailbreaking Leading Safety-Aligned LLMs with Simple Adaptive Attacks](https://mlsanity.com/paper/2404.02151/) (2024)
- [Safety Alignment Should Be Made More Than Just a Few Tokens Deep](https://mlsanity.com/paper/2406.05946/) (2024)
- [Safety-Tuned LLaMAs: Lessons From Improving the Safety of Large Language Models that Follow Instructions](https://mlsanity.com/paper/2309.07875/) (2023)
- [Improving Alignment and Robustness with Circuit Breakers](https://mlsanity.com/paper/2406.04313/) (2024)
- [SafeDecoding: Defending against Jailbreak Attacks via Safety-Aware Decoding](https://mlsanity.com/paper/2402.08983/) (2024)
- [JailBreakV: A Benchmark for Assessing the Robustness of MultiModal Large Language Models against Jailbreak Attacks](https://mlsanity.com/paper/2404.03027/) (2024)
- [VoiceBench: Benchmarking LLM-Based Voice Assistants](https://mlsanity.com/paper/2410.17196/) (2024)
- [Llama Guard 3 Vision: Safeguarding Human-AI Image Understanding Conversations](https://mlsanity.com/paper/2411.10414/) (2024)
- [Eyes Closed, Safety On: Protecting Multimodal LLMs via Image-to-Text Transformation](https://mlsanity.com/paper/2403.09572/) (2024)
- [MLLM-Protector: Ensuring MLLM's Safety without Hurting Performance](https://mlsanity.com/paper/2401.02906/) (2024)
- [GradSafe: Detecting Jailbreak Prompts for LLMs via Safety-Critical Gradient Analysis](https://mlsanity.com/paper/2402.13494/) (2024)
- [CodeAttack: Revealing Safety Generalization Challenges of Large Language Models via Code Completion](https://mlsanity.com/paper/2403.07865/) (2024)
- [Immune: Improving Safety Against Jailbreaks in Multi-modal LLMs via Inference-Time Alignment](https://mlsanity.com/paper/2411.18688/) (2024)
- [Audio Is the Achilles' Heel: Red Teaming Audio Large Multimodal Models](https://mlsanity.com/paper/2410.23861/) (2024)
- [SafeInfer: Context Adaptive Decoding Time Safety Alignment for Large Language Models](https://mlsanity.com/paper/2406.12274/) (2024)
- [TurboFuzzLLM: Turbocharging Mutation-based Fuzzing for Effectively Jailbreaking Large Language Models in Practice](https://mlsanity.com/paper/2502.18504/) (2025)

## Built on by (most cited first, 2 of 2)

- [SafeRI: Recognition and Intervention for Token-Level Safety Intervention in Large Vision Language Models](https://mlsanity.com/paper/2609.03544/) (2026)
- [Can Vision-Language Models Stay Helpful When Facing Implicit Risks? Intent-Privilege OPSD for Efficient Safety-Helpfulness Alignment](https://mlsanity.com/paper/2609.37837/) (2026)

## Abstract (arXiv)

Multimodal large-language models (MLLMs) often experience degraded safety alignment when harmful queries exploit cross-modal interactions. Models aligned on text alone show a higher rate of successful attacks when extended to two or more modalities. We propose a simple conditional decoding strategy, CASA (Classification Augmented with Safety Attention) that uses internal representations of MLLMs to predict a binary safety token before response generation. The classification is guided by a safety attention mechanism that computes attention between a dedicated query prompt and the input to scale the classification logits, sharpening the model's ability to detect malicious queries. Our design ensures robust safety alignment without relying on any external classifier or auxiliary head, and without modality-specific safety fine-tuning. Unlike prior safety-decoding methods that attach an external classifier or moderation head, or rescale every generated token with a reward model, CASA makes a single, internal, attention-gated safety decision over the shared multimodal representation, which is precisely what lets safety learned from text-only training transfer to image and audio inputs. Compared against 4 baseline models across 7 attack types spanning text, image, and audio modalities, evaluated on benchmarks such as MM-SafetyBench, JailbreakV-28k, and adversarial audio tests, CASA lowers the average attack success rate by >90% relative when paired with a standard safety prompt (applied to all baselines for fair comparison), and by 54--61% using its internal safety mechanism alone. Our empirical evaluations also show that CASA~maintains strong utility on benign inputs, validated through both automated and human evaluations (via 13 trained annotators). Together, these results highlight CASA as a simple and generalizable framework to improve multimodal LLM safety.