- Rating
- 3
- Confidence
- 4
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
The paper adapts sampling-based LLM watermarking to propose a black-box, key-averaged distortion-free semantic watermark, applied to text-to-image models. The robustness is achieved by the use of a semantic feature extractor (plus LSH) which is robust to content-preserving image manipulation, while distortion-freeness comes from grouping candidates by LSH code and a keyed selection rule that preserves the probability. In a nutshell, this is a zero-bit watermarking method (no payload, but a statistical binary detection problem), while for image generation models, an important branch is multi-bit methods (with payload, need to extract the watermark and calculate bit error rate, e.g., Gaussian Shading, but with white-box requirement). With the advantage of only requiring black-box access, one price paid is generating candidate outputs per released image (and a larger does not necessarily guarantee a better result), while another price or limitation is weak single-image detection. The authors should discuss whether a multi-bit extension is feasible and at what cost in .
Critical strengths
- The paper is clearly written and easy to follow.
- The idea is new in the image generation regime.
- The theoretical justification and experiments are adequate to support the claim.
- Limitations are honestly stated (single image, adaptive attack).
Critical weaknesses
-
The core mechanism combines bundle sampling, black-box sample-based watermarking, and LSH. Beyond the change of modality and a fine-tuned encoder, the technical difference is unclear.
-
The threat model is unclear, which leads to uncertainty on how the method is deployed (see question 2 for detail).
-
The scheme is zero-bit (no payload), single-image detection is weak, and each image costs generations.
Questions for authors
-
The method requires discretizing the semantic embedding into codes, and the authors adopt LSH without justifying it over other binarization alternatives (there are many). This choice will potentially affect Theorem 4 and it is worth ablation.
-
The threat model is not clarified. The paper does not state who deploys the scheme (model owner, API service, or end user), who holds the key , who performs detection, or whether is public or secret. If the owner deploys it, it makes much less sense to treat the model as a black box. Instead, the owner should have white-box access and be motivated to leverage it to design a stronger watermark with the best fidelity, rather than restricting itself to black-box access in the first place.
-
The authors should discuss whether a multi-bit extension is feasible and at what cost in .
Thank you for the detailed rebuttal. I find the main contribution interesting. The rebuttal has addressed most of my concerns and made the contribution more convincing. I still think adaptive robustness is an important limitation, but I appreciate that the authors acknowledge it.
Overall, I view the paper more positively now and am happy to raise my score.