AI Paper Atlas

Research map / Large language models / Instruction and fine-tuning

Safety Mirage: How Spurious Correlations Undermine VLM Safety Fine-Tuning and Can Be Mitigated by Machine Unlearning

Yiwei Chen, Yuguang Yao, Yihua Zhang, Bingquan Shen, Gaowen Liu, Sijia Liu

ICLR 2026 · Large language models › Instruction and fine-tuning · approach: Safety alignment after fine-tuning · 15 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace Mixed-SFT safety fine-tuning with NPO unlearningLLaVA-1.5-7B · VLGuardASR after one-word attack: 54.98% -> 12.92%Improved
replace Mixed-SFT with RMU unlearningLLaVA-1.5-7B · VLGuardASR after one-word attack: 54.98% -> 10.18%Improved
replace Posthoc-SFT with RMU unlearningLLaVA-1.5-7B · SPA-VLASR after one-word attack: 32.96% -> 22.64%Improved
replace Mixed-SFT with NPO unlearning to reduce over-prudenceLLaVA-1.5-7B · VLGuardrefusal rate on benign queries: 91.76% -> 11.69%Improved
replace Posthoc-SFT with RMU unlearning to reduce over-prudenceLLaVA-1.5-7B · SPA-VLrefusal rate on benign queries: 100.0% -> 19.24%Improved
add Unsafe-Filter data filtering to unsafe base modelLLaVA-1.5-7B · VLGuardASR after one-word attack: 90.27% -> 90.72%No clear effect
replace SFT with unlearning, utility costLLaVA-1.5-7B · VQAv2accuracy: 78.23% -> 77.04%Mixed
use unlearning with 10% of the forget data (NPO) instead of full (ablation)LLaVA-1.5-7B · VLGuardASR after one-word attack: 12.92% -> 18.77%Hurt
use unlearning with 10% of the forget data (RMU) instead of full (ablation)LLaVA-1.5-7B · VLGuardASR after one-word attack: 10.18% -> 22.12%Hurt
add Gaussian noise to images under Mixed-SFT (ablation)LLaVA-1.5-7B · VLGuardASR after one-word attack: 54.98% -> 54.08%No clear effect
apply NPO unlearning rather than SFT under GCG attackLLaVA-1.5-7B · VLGuardASR after GCG attack: 72.90% -> 3.87%Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy