Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

POPS: Recovering Unlearned Multi-Modality Knowledge in MLLMs with Prompt-Optimized Parameter Shaking

About

Multimodal Large Language Models (MLLMs) have demonstrated impressive performance on cross-modal tasks by jointly training on large-scale textual and visual data, where privacy-sensitive examples could be unintentionally encoded, raising concerns about privacy or copyright violation. To this end, Multi-modality Machine Unlearning (MMU) was proposed as a mitigation that can effectively force MLLMs to forget private information. However, the robustness of such unlearning methods is not fully exploited when the model is published and accessible to malicious users. In this paper, we propose a novel adversarial strategy, namely Prompt-Optimized Parameter Shaking (POPS), aiming to recover the supposedly unlearned multi-modality knowledge from the MLLMs. Our method elicits the victim MLLMs to generate potential private examples via prompt-suffix optimization, and then exploits these synthesized outputs to fine-tune the models so they disclose the true private information. The experiments on the different MMU benchmarks reveal substantial weaknesses in the existing MMU algorithms. Our POPS can even achieve a near-complete recovery of supposedly erased sensitive information on the unlearned MLLMs, exposing fundamental vulnerabilities that challenge the foundational robustness of representative MMU-based privacy protections.

Zhangheng LI, Jianing Zhu, Junyuan Hong, Sungmin Eum, Shuowen Hu, Suya You, Zhangyang Wang• 2026

Related benchmarks

TaskDatasetResultRank
Multimodal Machine Unlearning EvaluationMLLMU-Bench (test)
Classification Accuracy43.05
34
Machine UnlearningMLLMU-Bench (Retain Set)
Accuracy44.42
7
Adversarial Knowledge RecoveryMLLMU-Bench (test)
Test Accuracy42.9
4
Showing 3 of 3 rows

Other info

Follow for update