Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2602.18845 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915896088330240 |
|---|---|
| author | Xia, Chengwei Ma, Fan Quan, Ruijie Xu, Yunqiu Zhan, Kun Yang, Yi |
| author_facet | Xia, Chengwei Ma, Fan Quan, Ruijie Xu, Yunqiu Zhan, Kun Yang, Yi |
| contents | With the rapid deployment of multimodal large language models (MLLMs), disputes regarding model ownership have become increasingly frequent, raising significant concerns about intellectual property protection. In this paper, we propose a framework for generating copyright triggers for MLLMs, enabling model publishers to embed verifiable ownership information into the model. The goal is to construct trigger images that elicit ownership-related textual responses exclusively in fine-tuned derivatives, while remaining inert in other non-derivative models. Our method constructs a tracking trigger image by treating the image as a learnable tensor, performing adversarial optimization with dual-injection of ownership-relevant semantic information. The first injection is achieved by enforcing textual consistency between the output of an auxiliary MLLM and a predefined ownership-relevant target text; the consistency loss is backpropagated to inject this ownership-related information into the image. The second injection is performed at the semantic-level by minimizing the distance between the CLIP features of the image and those of the target text. Furthermore, we introduce an additional adversarial training stage involving the auxiliary model. It is specifically trained to resist generating ownership-relevant target text, thereby enhancing robustness in heavily fine-tuned derivative models. Extensive experiments demonstrate the effectiveness of our dual-injection approach in tracking model lineage under various fine-tuning and domain-shift scenarios. Code is at https://github.com/kunzhan/AGDI |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_18845 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Echoes of ownership: Adversarial-guided dual injection for copyright protection in MLLMs Xia, Chengwei Ma, Fan Quan, Ruijie Xu, Yunqiu Zhan, Kun Yang, Yi Computer Vision and Pattern Recognition With the rapid deployment of multimodal large language models (MLLMs), disputes regarding model ownership have become increasingly frequent, raising significant concerns about intellectual property protection. In this paper, we propose a framework for generating copyright triggers for MLLMs, enabling model publishers to embed verifiable ownership information into the model. The goal is to construct trigger images that elicit ownership-related textual responses exclusively in fine-tuned derivatives, while remaining inert in other non-derivative models. Our method constructs a tracking trigger image by treating the image as a learnable tensor, performing adversarial optimization with dual-injection of ownership-relevant semantic information. The first injection is achieved by enforcing textual consistency between the output of an auxiliary MLLM and a predefined ownership-relevant target text; the consistency loss is backpropagated to inject this ownership-related information into the image. The second injection is performed at the semantic-level by minimizing the distance between the CLIP features of the image and those of the target text. Furthermore, we introduce an additional adversarial training stage involving the auxiliary model. It is specifically trained to resist generating ownership-relevant target text, thereby enhancing robustness in heavily fine-tuned derivative models. Extensive experiments demonstrate the effectiveness of our dual-injection approach in tracking model lineage under various fine-tuning and domain-shift scenarios. Code is at https://github.com/kunzhan/AGDI |
| title | Echoes of ownership: Adversarial-guided dual injection for copyright protection in MLLMs |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2602.18845 |