# Your Text Encoder Can Be An Object-Level Watermarking Controller

arXiv: https://arxiv.org/abs/2503.11945 · HTML: https://mlsanity.com/paper/2503.11945/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Watermarking generated content and models, one of 12 directions in LLM safety and security.
- Builds on 26 of its 41 references that are in the map.
- First posted on arXiv 15 Mar 2025.

## Placed in

- LLM safety and security / [Watermarking generated content and models](https://mlsanity.com/safety/watermarking-generated-content-and-models/)

## Builds on (most cited first, 26 of 26)

- [High-Resolution Image Synthesis with Latent Diffusion Models](https://mlsanity.com/paper/2112.10752/) (2021)
- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [Denoising Diffusion Implicit Models](https://mlsanity.com/paper/2010.02502/) (2020)
- [An Image is Worth One Word: Personalizing Text-to-Image Generation using Textual Inversion](https://mlsanity.com/paper/2208.01618/) (2022)
- [Diffusion Models for Adversarial Purification](https://mlsanity.com/paper/2205.07460/) (2022)
- [HiDDeN: Hiding Data With Deep Networks](https://mlsanity.com/paper/1807.09937/) (2018)
- [Rethinking FID: Towards a Better Evaluation Metric for Image Generation](https://mlsanity.com/paper/2401.09603/) (2023)
- [The Stable Signature: Rooting Watermarks in Latent Diffusion Models](https://mlsanity.com/paper/2303.15435/) (2023)
- [Inversion-Based Style Transfer with Diffusion Models](https://mlsanity.com/paper/2211.13203/) (2022)
- [Invisible Image Watermarks Are Provably Removable Using Generative AI](https://mlsanity.com/paper/2306.01953/) (2023)
- [Soft Diffusion: Score Matching for General Corruptions](https://mlsanity.com/paper/2209.05442/) (2022)
- [Diffusion4D: Fast Spatial-temporal Consistent 4D Generation via Video Diffusion Models](https://mlsanity.com/paper/2405.16645/) (2024)
- [WAVES: Benchmarking the Robustness of Image Watermarks](https://mlsanity.com/paper/2401.08573/) (2024)
- [A survey of deep neural network watermarking techniques](https://mlsanity.com/paper/2103.09274/) (2021)
- [RoSteALS: Robust Steganography using Autoencoder Latent Space](https://mlsanity.com/paper/2304.03400/) (2023)
- [Watermark Anything with Localized Messages](https://mlsanity.com/paper/2411.07231/) (2024)
- [RB-Modulation: Training-Free Personalization of Diffusion Models using Stochastic Optimal Control](https://mlsanity.com/paper/2405.17401/) (2024)
- [AquaLoRA: Toward White-box Protection for Customized Stable Diffusion Models via Watermark LoRA](https://mlsanity.com/paper/2405.11135/) (2024)
- [A Watermark-Conditioned Diffusion Model for IP Protection](https://mlsanity.com/paper/2403.10893/) (2024)
- [LaWa: Using Latent Space for In-Generation Image Watermarking](https://mlsanity.com/paper/2408.05868/) (2024)
- [Catch You Everything Everywhere: Guarding Textual Inversion via Concept Watermarking](https://mlsanity.com/paper/2309.05940/) (2023)
- [Scene-Conditional 3D Object Stylization and Composition](https://mlsanity.com/paper/2312.12419/) (2023)
- [Perception-oriented Single Image Super-Resolution via Dual Relativistic Average Generative Adversarial Networks](https://mlsanity.com/paper/1911.03464/) (2019)
- [Survival Analysis of Young Triple-Negative Breast Cancer Patients](https://mlsanity.com/paper/2401.08712/) (2024)
- [SSyncOA: Self-synchronizing Object-aligned Watermarking to Resist Cropping-paste Attacks](https://mlsanity.com/paper/2405.03458/) (2024)
- [Exploring Iterative Manifold Constraint for Zero-shot Image Editing](https://mlsanity.com/paper/2501.03631/) (2025)

## Built on by (most cited first, 1 of 1)

- [Score-Control for Hallucination Reduction in Diffusion Models](https://mlsanity.com/paper/2606.00377/) (2026)

## Abstract (arXiv)

Invisible watermarking of AI-generated images can help with copyright protection, enabling detection and identification of AI-generated media. In this work, we present a novel approach to watermark images of T2I Latent Diffusion Models (LDMs). By only fine-tuning text token embeddings W_*, we enable watermarking in selected objects or parts of the image, offering greater flexibility compared to traditional full-image watermarking. Our method leverages the text encoder's compatibility across various LDMs, allowing plug-and-play integration for different LDMs. Moreover, introducing the watermark early in the encoding stage improves robustness to adversarial perturbations in later stages of the pipeline. Our approach achieves 99% bit accuracy (48 bits) with a 105 × reduction in model parameters, enabling efficient watermarking.