Pith. sign in

REVIEW 3 major objections 2 minor 87 references

Memory Enhanced Fractional-Order Dung Beetle Optimization for Photovoltaic Parameter Identification

T0 review · 3 major / 2 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper's abstract proposes MFO-DBO, a fractional-order Dung Beetle Optimizer it claims wins photovoltaic parameter identification benchmarks, but the provided body is an unrelated paper on stealing prompts from text-to-image models.

desk verdict This submission is a broken artifact: the abstract describes a DBO variant for PV identification, but the full text is an unrelated paper on prompt stealing, so the claimed results simply do not exist here. read the letter →

arxiv 2508.06841 v1 pith:CXRFIHIS submitted 2025-08-09 cs.NE

Yiwei Li , Zhihua Allen-Zhao , Yuncheng Xu , Sanyang Liu This is my paper
classification cs.NE MSC 90C59
keywords fractional-orderoptimizationdungbeetleoptimizerphotovoltaicparameteridentificationchaoticmapinitializationCEC2017benchmarkmetaheuristicsmanuscriptbodymismatchpromptstealing(bodycontent)
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper, as represented by its abstract, proposes Memory Enhanced Fractional-Order Dung Beetle Optimization (MFO-DBO), an optimizer that grafts fractional-order calculus onto Dung Beetle Optimization, seeds the population with a fractional-order logistic chaotic map, and uses chaotic perturbation to help elite solutions escape local optima. The abstract claims that on the CEC2017 benchmark suite and the photovoltaic parameter identification problem, MFO-DBO consistently outperforms advanced DBO variants, CEC competition winners, fractional-order optimizers, enhanced classical algorithms, and recent metaheuristics in accuracy, robustness, and convergence speed. The provided manuscript body, however, is a complete, unrelated paper on stealing prompts from text-to-image diffusion models, with a different title, author list, and arXiv number; it never mentions MFO-DBO, Dung Beetle Optimization, fractional-order calculus, photovoltaic models, or the CEC2017 suite. A careful reader therefore cannot find the algorithm's specification, pseudocode, or the benchmark tables on which the abstract's claims would rest. The abstract's central claim is, in this submission, unsupported by the accompanying text.

What carries the argument

The claimed machinery is a triad: fractional-order calculus injected into the DBO update rules to give search steps a memory effect and smoother convergence; a fractional-order logistic chaotic map to seed the population with diversity; and a chaotic perturbation applied to elite solutions to escape local optima. The abstract asserts that these three components jointly deliver the reported accuracy, robustness, and speed gains, but the body provides no equations, pseudocode, or experimental results for any of them.

What would settle it

Open the provided manuscript and search the body for the strings 'Dung Beetle', 'fractional-order', 'photovoltaic', and 'CEC2017'; the complete absence of these terms — with the body instead containing the full text of a prompt-stealing paper under a different title and author list — settles whether the abstract's claims are supported. Conversely, locating the promised algorithm description, pseudocode, and benchmark tables in a complete body would refute this assessment.

Watch

Extended reading notes

Core claim

The paper's claim, on its own terms, is that a metaheuristic called MFO-DBO — Dung Beetle Optimization augmented with fractional-order memory, a fractional-order logistic chaotic map for population initialization, and a chaotic perturbation mechanism for elite individuals — solves the nonlinear, multimodal, high-dimensional photovoltaic parameter identification problem and outperforms a wide field of modern optimizers on the CEC2017 suite and PV models, with a better exploration–exploitation balance than standard DBO. The body of this submission, however, presents a different paper entirely: a training-free, proxy-in-the-loop prompt-stealing attack that recovers text-to-image prompts by gene

Load-bearing premise

The load-bearing premise is that the manuscript body actually describes MFO-DBO; on inspection, the body is a complete unrelated paper, so the abstract's algorithm, benchmarks, and conclusions have no supporting text in this submission.

Editorial extensions

If this is right

  • If the abstract's claims are credited, the combination of fractional-order memory and chaotic-map initialization would give PV parameter identification a solver that is more diverse in exploration and more stable near convergence, directly targeting the premature-convergence weakness of standard DBO.
  • Fractional-order chaotic initialization, if it works as claimed, would be a drop-in replacement for the seeding step of other swarm algorithms, not just DBO.
  • The chaotic perturbation on elite solutions would, on the paper's account, provide a general local-optima escape mechanism usable in any elitist metaheuristic.
  • Demonstrated success on CEC2017, as claimed, would position MFO-DBO as a general-purpose black-box optimizer beyond photovoltaics, since that suite is a standard testbed for such comparisons.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The discrepancy between the abstract and the body is the decisive fact about this submission: the abstract's evaluation claims — CEC2017 tables, PV fitting results, comparisons against named baselines — cannot be verified because the accompanying text describes a different paper; verification would require the missing experimental section.
  • If a complete MFO-DBO manuscript later appears, the component to examine first is the fractional-order memory mechanism, since the claimed advantage over prior DBO variants rests on that mechanism rather than on chaotic perturbations, which are a common add-on in swarm optimizers.
  • A testable extension suggested by the abstract alone: implement DBO with and without the fractional-order memory term, holding the chaotic initialization fixed, and compare the two on a mix of unimodal and multimodal CEC2017 functions to isolate the memory component's contribution.
  • Read at face value, the body is itself a coherent paper about prompt-stealing attacks; a reader evaluating that content would be assessing a different set of claims, meaning the submission is in effect two documents with mismatched fronts.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 2 minor

Summary. The submission claims, in its title and abstract, to propose a Memory Enhanced Fractional-Order Dung Beetle Optimization (MFO-DBO) algorithm for photovoltaic (PV) parameter identification, integrating fractional-order calculus memory, a fractional-order logistic chaotic map, and a chaotic perturbation mechanism, and to demonstrate consistent superiority over multiple optimizer families on CEC2017 and PV benchmark problems. However, the full text supplied under this title is an entirely different paper: "Towards Effective Prompt Stealing Attack against Text-to-Image Diffusion Models," arXiv:2508.06837v2, authored by a different team and carrying an NDSS 2026 header. The body contains no occurrence of MFO-DBO, DBO, fractional-order, CEC2017, or photovoltaic parameter identification. Its algorithms are GreedyProxyQuery and Dynamic Modifier Extraction, and its tables report CLIP, LPIPS, SBERT, and ASR metrics for prompt-stealing performance. None of the claimed algorithm specification, equations, pseudocode, benchmark comparisons, or PV parameter identification results appear in the submitted text.

Significance. The significance of the claimed result cannot be assessed from the submitted artifact. A paper whose abstract promises a new optimizer with benchmark validation must, at minimum, define the optimizer and report its results; this submission does neither. The body is a complete, unrelated paper with its own arXiv number and author list. Consequently, the central claim of the abstract has no supporting derivation, no reproducibility basis, and no evidence within the manuscript. There is no positive contribution that can be evaluated on the merits.

major comments (3)
  1. [Full text, overall] The full text is a different paper, 'Towards Effective Prompt Stealing Attack against Text-to-Image Diffusion Models' (arXiv:2508.06837v2, NDSS 2026), with a different author list. The submitted title and abstract describe MFO-DBO, but no passage in the body uses the terms MFO-DBO, dung beetle, fractional-order, CEC2017, or photovoltaic. This is not a missing auxiliary file or a local omission; the evaluated object described in the abstract is absent from the manuscript.
  2. [Full text, Sections IV and V] The central algorithmic claims of the abstract are undefined. There is no equation defining the fractional-order memory mechanism, no specification of the fractional-order logistic chaotic map, no chaotic perturbation update, and no pseudocode for MFO-DBO. The only algorithms present are Algorithm 1 (GreedyProxyQuery) and Algorithm 2 (Dynamic Modifier Extraction), which belong to the prompt-stealing paper. Consequently, no reviewer can check the claimed improvements in accuracy, robustness, or convergence speed, nor can the method be reproduced.
  3. [Full text, Section V (tables)] No benchmark results for the claimed claims exist. The abstract promises numerical results on CEC2017 and PV parameter identification, but all tables in the body report prompt-stealing metrics (CLIP img, LPIPS, SBERT, ASR) across FLUX, ShuttleDiffusion, and Stable Diffusion-3.5. There is no comparison with DBO variants, CEC competition winners, FO-based optimizers, enhanced classical algorithms, or recent metaheuristics. The abstract's performance claims are therefore unsupported by any data in the submission.
minor comments (2)
  1. [Header and metadata] The manuscript metadata is internally inconsistent: the submitted arXiv number is 2508.06841, while the body's header states arXiv:2508.06837v2 [cs.CR] and an NDSS 2026 copyright line. The author list also differs from what the abstract implies. These inconsistencies should be resolved if a corrected submission is made.
  2. [Title and abstract] The title and abstract do not correspond to the body text. Even as a formatting or packaging error, this prevents the standard review process from operating on the claimed contribution.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the provided full text contains no MFO-DBO derivation, so no claimed result can be exhibited as reducing to its own inputs by construction.

full rationale

The submission's abstract describes MFO-DBO, a fractional-order dung beetle optimizer, and claims results on CEC2017 and photovoltaic parameter identification. The FULL TEXT, however, is an unrelated paper, arXiv:2508.06837v2, 'Towards Effective Prompt Stealing Attack against Text-to-Image Diffusion Models', by a different author team. The body specifies Prometheus, not MFO-DBO: Algorithm 1 is GreedyProxyQuery and Algorithm 2 is Dynamic Modifier Extraction, with tables reporting CLIP, LPIPS, SBERT, and ASR metrics. No equation or pseudocode in the provided text defines the fractional-order memory mechanism, the fractional-order logistic chaotic map, or the chaotic perturbation mechanism, and no benchmark results for CEC2017 or PV parameter identification appear. The circularity rubric requires exhibiting a specific reduction by construction (e.g., a fitted parameter renamed as a prediction, or a self-citation chain that forces the conclusion). Here there is no derivation chain to audit at all: the abstract's claims are unsupported by the artifact, but unsupported is not circular. The appropriate finding is therefore no significant circularity, with the absent derivation noted as a correctness/integrity problem outside this axis.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

The artifact provides essentially nothing for the claimed MFO-DBO contribution: no pseudocode, no hyperparameter values, no benchmark tables, no code repository. The ledger records the assumptions under which the abstract's assertions could be true, all unverifiable because the body is a different paper. No invented entities are introduced by the claimed algorithm in the available text.

free parameters (1)
  • MFO-DBO hyperparameters (fractional order, chaotic map coefficients, perturbation strength, memory size) = not reported in provided text
    The abstract attributes gains to FO calculus, a fractional-order logistic chaotic map, and chaotic perturbation; no parameter values or sensitivity analysis appear because the algorithm itself is absent from the body.
assumptions (3)
  • domain assumption The manuscript body corresponds to the abstract's claimed algorithm
    The review must assume the text under FULL TEXT is the paper being submitted; this assumption fails, since the body is an unrelated prompt-stealing paper with its own arXiv number.
  • domain assumption CEC2017 and PV parameter identification experiments were actually run and support the abstract's claims
    No such experiments exist in the provided text; the abstract's performance assertions rest on results that are not present.
  • standard math Standard metaheuristic evaluation protocol (mean/std over runs, statistical tests)
    Typical for such claims, but unverifiable here; nothing in the body reports run-level statistics for MFO-DBO.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Memory Enhanced Fractional-Order Dung Beetle Optimization for Photovoltaic Parameter Identification." pith.science (2026). https://pith.science/paper/CXRFIHIS

@misc{pith2026250806841,
  author       = {Pith},
  title        = {Pith review of: Memory Enhanced Fractional-Order Dung Beetle Optimization for Photovoltaic Parameter Identification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CXRFIHIS}},
  note         = {Machine review of arXiv:2508.06841}
}
read the original abstract

Accurate parameter identification in photovoltaic (PV) models is crucial for performance evaluation but remains challenging due to their nonlinear, multimodal, and high-dimensional nature. Although the Dung Beetle Optimization (DBO) algorithm has shown potential in addressing such problems, it often suffers from premature convergence. To overcome these issues, this paper proposes a Memory Enhanced Fractional-Order Dung Beetle Optimization (MFO-DBO) algorithm that integrates three coordinated strategies. Firstly, fractional-order (FO) calculus introduces memory into the search process, enhancing convergence stability and solution quality. Secondly, a fractional-order logistic chaotic map improves population diversity during initialization. Thirdly, a chaotic perturbation mechanism helps elite solutions escape local optima. Numerical results on the CEC2017 benchmark suite and the PV parameter identification problem demonstrate that MFO-DBO consistently outperforms advanced DBO variants, CEC competition winners, FO-based optimizers, enhanced classical algorithms, and recent metaheuristics in terms of accuracy, robustness, convergence speed, while also maintaining an excellent balance between exploration and exploitation compared to the standard DBO algorithm.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

87 extracted references · 51 canonical work pages

  1. [1]

    Recruitment of prompt engineer

    Adobe. Recruitment of prompt engineer. https://careers.adobe.com/ us/en/job/R158292/Prompt-Engineer?utm source=chatgpt.com, 2025. Accessed on: 2025-10-26

    2025

  2. [2]

    Leonardo ai

    Leonardo AI. Leonardo ai. https://app.leonardo.ai/, 2024. Accessed on: 2024-06-26

    2024

  3. [3]

    stable-diffusion-3.5-large-turbo

    Stability AI. stable-diffusion-3.5-large-turbo. https://huggingface.co/stabi lityai/stable-diffusion-3.5-large-turbo, 2024. Accessed on: 2024-11-28

    2024

  4. [4]

    AIFrog. Aifrog. https://www.aifrog.io/, 2024. Accessed on: 2024-06-26

    2024

  5. [5]

    Extracting training data from diffusion models

    Nicolas Carlini, Jamie Hayes, Milad Nasr, Matthew Jagielski, Vikash Sehwag, Florian Tramer, Borja Balle, Daphne Ippolito, and Eric Wallace. Extracting training data from diffusion models. In 32nd USENIX Security Symposium (USENIX Security 23), pages 5253–5270, 2023

    2023

  6. [6]

    Recruitment of prompt engineer

    Delphi. Recruitment of prompt engineer. https://www.indeed.com/q-pro mpt-engineer-jobs.html?utm source=chatgpt.com&vjk=576c20b3cfe4 76dd, 2025. Accessed on: 2025-10-26

    2025

  7. [7]

    Vivid bright bold graphic illustrations

    Dianamoran@PromptBase. Vivid bright bold graphic illustrations. https: //promptbase.com/prompt/vivid-bright-bold-graphic-illustrations-2, 2024. Accessed on: 2024-06-26

    2024

  8. [8]

    Are diffusion models vulnerable to membership inference attacks? In International Conference on Machine Learning, pages 8717–8730

    Jinhao Duan, Fei Kong, Shiqi Wang, Xiaoshuang Shi, and Kaidi Xu. Are diffusion models vulnerable to membership inference attacks? In International Conference on Machine Learning, pages 8717–8730. PMLR, 2023

    work page 2023

Show all 87 references
  1. [9]

    Bae: Bert-based adversarial examples for text classification

    Siddhant Garg and Goutham Ramakrishnan. Bae: Bert-based adversarial examples for text classification. arXiv preprint arXiv:2004.01970, 2020

    2004 arXiv

  2. [10]

    Copyright- shield: Enhancing diffusion model security against copyright infringement attacks

    Zhixiang Guo, Siyuan Liang, Aishan Liu, and Dacheng Tao. Copyright- shield: Enhancing diffusion model security against copyright infringement attacks. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 19417–19426, 2025

    2025

  3. [11]

    Scalable zero-knowledge proofs for non-linear functions in machine learning

    Meng Hao, Hanxiao Chen, Hongwei Li, Chenkai Weng, Yuan Zhang, Haomiao Yang, and Tianwei Zhang. Scalable zero-knowledge proofs for non-linear functions in machine learning. In 33rd USENIX Security Symposium (USENIX Security 24), pages 3819–3836, 2024

    2024

  4. [12]

    Iron: Private inference on transformers

    Meng Hao, Hongwei Li, Hanxiao Chen, Pengzhi Xing, Guowen Xu, and Tianwei Zhang. Iron: Private inference on transformers. Advances in neural information processing systems, 35:15718–15731, 2022

    2022

  5. [13]

    Optimizing prompts for text-to-image generation

    Yaru Hao, Zewen Chi, Li Dong, and Furu Wei. Optimizing prompts for text-to-image generation. Advances in Neural Information Processing Systems, 36, 2024

    2024

  6. [14]

    Dall-e-prompts

    Sharath S Hebbar. Dall-e-prompts. https://huggingface.co/datasets/Sharat hhebbar24/DALL-E-Prompts-OpenAI-ChatGPT?row=4, 2024. Accessed on: 2024-11-28

    2024

  7. [15]

    Denoising diffusion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffusion probabilistic models. Advances in neural information processing systems, 33:6840–6851, 2020

    2020

  8. [16]

    Membership inference of diffusion models

    Hailong Hu and Jun Pang. Membership inference of diffusion models. arXiv preprint arXiv:2301.09956, 2023

    2023 arXiv

  9. [17]

    Enhancing adversarial example transferability with an intermediate level attack

    Qian Huang, Isay Katsman, Horace He, Zeqi Gu, Serge Belongie, and Ser-Nam Lim. Enhancing adversarial example transferability with an intermediate level attack. In Proceedings of the IEEE/CVF international conference on computer vision, pages 4733–4742, 2019

    2019

  10. [18]

    Personalization as a shortcut for few-shot backdoor attack against text-to-image diffusion models

    Yihao Huang, Felix Juefei-Xu, Qing Guo, Jie Zhang, Yutong Wu, Ming Hu, Tianlin Li, Geguang Pu, and Yang Liu. Personalization as a shortcut for few-shot backdoor attack against text-to-image diffusion models. In Proceedings of the AAAI Conference on Artificial Intelligence, vol...

    2024

  11. [19]

    Pleak: Prompt leaking attacks against large language model applications

    Bo Hui, Haolin Yuan, Neil Gong, Philippe Burlina, and Yinzhi Cao. Pleak: Prompt leaking attacks against large language model applications. In Proceedings of the 2024 on ACM SIGSAC Conference on Computer and Communications Security, pages 3600–3614, 2024

    2024

  12. [20]

    Gpt-4o system card

    Aaron Hurst, Adam Lerer, Adam P Goucher, Adam Perelman, Aditya Ramesh, Aidan Clark, AJ Ostrow, Akila Welihinda, Alan Hayes, Alec Radford, et al. Gpt-4o system card. arXiv preprint arXiv:2410.21276, 2024

    2024 arXiv

  13. [21]

    Promptsafe: Gated prompt tuning for safe text-to-image generation

    Zonglei Jing, Xiao Yang, Xiaoqian Li, Siyuan Liang, Aishan Liu, Mingchuan Zhang, and Xianglong Liu. Promptsafe: Gated prompt tuning for safe text-to-image generation. arXiv preprint arXiv:2508.01272, 2025

    2025 arXiv

  14. [22]

    Adversarial machine learning at scale

    Alexey Kurakin, Ian Goodfellow, and Samy Bengio. Adversarial machine learning at scale. arXiv preprint arXiv:1611.01236, 2016

    2016 arXiv

  15. [23]

    Flux.1-schnell

    Black Forest Labs. Flux.1-schnell. https://huggingface.co/black-forest-l abs/FLUX.1-schnell, 2024. Accessed on: 2024-11-28

    2024

  16. [24]

    Lexica dataset

    Lexica. Lexica dataset. https://lexica.art/, 2024. Accessed on: 2024-06-26

    2024

  17. [25]

    Dream: Scalable red teaming for text-to-image generative systems via distribution modeling

    Boheng Li, Junjie Wang, Yiming Li, Zhiyang Hu, Leyi Qi, Jianshuo Dong, Run Wang, Han Qiu, Zhan Qin, and Tianwei Zhang. Dream: Scalable red teaming for text-to-image generative systems via distribution modeling. arXiv preprint arXiv:2507.16329, 2025

    2025

  18. [26]

    Textbugger: Generating adversarial text against real-world applications

    Jinfeng Li, Shouling Ji, Tianyu Du, Bo Li, and Ting Wang. Textbugger: Generating adversarial text against real-world applications. arXiv preprint arXiv:1812.05271, 2018

    2018 arXiv

  19. [27]

    Blip: Bootstrapping language-image pre-training for unified vision-language understanding and generation

    Junnan Li, Dongxu Li, Caiming Xiong, and Steven Hoi. Blip: Bootstrapping language-image pre-training for unified vision-language understanding and generation. In International conference on machine learning, pages 12888–12900. PMLR, 2022

    2022

  20. [28]

    Rethinking data protection in the (generative) artificial intelligence era

    Yiming Li, Shuo Shao, Yu He, Junfeng Guo, Tianwei Zhang, Zhan Qin, Pin-Yu Chen, Michael Backes, Philip Torr, Dacheng Tao, et al. Rethinking data protection in the (generative) artificial intelligence era. arXiv preprint arXiv:2507.03034, 2025. 15

    2025 arXiv

  21. [29]

    Safemobile: Chain-level jailbreak detection and automated evaluation for multimodal mobile agents

    Siyuan Liang, Tianmeng Fang, Zhe Liu, Aishan Liu, Yan Xiao, Jinyuan He, Ee-Chien Chang, and Xiaochun Cao. Safemobile: Chain-level jailbreak detection and automated evaluation for multimodal mobile agents. arXiv preprint arXiv:2507.00841, 2025

    2025 arXiv

  22. [30]

    Revisiting backdoor attacks against large vision-language models from domain shift

    Siyuan Liang, Jiawei Liang, Tianyu Pang, Chao Du, Aishan Liu, Mingli Zhu, Xiaochun Cao, and Dacheng Tao. Revisiting backdoor attacks against large vision-language models from domain shift. In Proceedings of the Computer Vision and Pattern Recognition Conference, pages 9477– 9486, 2025

    2025

  23. [31]

    T2vshield: Model- agnostic jailbreak defense for text-to-video models

    Siyuan Liang, Jiayang Liu, Jiecheng Zhai, Tianmeng Fang, Rongcheng Tu, Aishan Liu, Xiaochun Cao, and Dacheng Tao. T2vshield: Model- agnostic jailbreak defense for text-to-video models. arXiv preprint arXiv:2504.15512, 2025

    2025 arXiv

  24. [32]

    Badclip: Dual-embedding guided backdoor attack on multimodal contrastive learning

    Siyuan Liang, Mingli Zhu, Aishan Liu, Baoyuan Wu, Xiaochun Cao, and Ee-Chien Chang. Badclip: Dual-embedding guided backdoor attack on multimodal contrastive learning. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 24645– 24654, 2024

    2024

  25. [33]

    Microsoft coco: Common objects in context

    Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Doll ´ar, and C Lawrence Zitnick. Microsoft coco: Common objects in context. In Computer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedin...

    2014

  26. [34]

    Microsoft coco: Common objects in context

    Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Doll ´ar, and C Lawrence Zitnick. Microsoft coco: Common objects in context. In Computer vision–ECCV 2014: 13th European conference, zurich, Switzerland, September 6-12, 2014, proceedin...

    2014

  27. [35]

    T2v-optjail: Discrete prompt optimization for text-to-video jailbreak attacks

    Jiayang Liu, Siyuan Liang, Shiqian Zhao, Rongcheng Tu, Wenbo Zhou, Aishan Liu, Dacheng Tao, and Siew Kei Lam. T2v-optjail: Discrete prompt optimization for text-to-video jailbreak attacks. arXiv preprint arXiv:2505.06679, 2025

    2025 arXiv

  28. [36]

    Design guidelines for prompt engineering text-to-image generative models

    Vivian Liu and Lydia B Chilton. Design guidelines for prompt engineering text-to-image generative models. In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems, pages 1–23, 2022

    2022

  29. [37]

    Prompting hard or hardly prompting: Prompt inversion for text-to-image diffusion models

    Shweta Mahajan, Tanzila Rahman, Kwang Moo Yi, and Leonid Sigal. Prompting hard or hardly prompting: Prompt inversion for text-to-image diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 6808–6817, 2024

    2024

  30. [38]

    Black box adversarial prompting for foundation models

    Natalie Maus, Patrick Chao, Eric Wong, and Jacob Gardner. Black box adversarial prompting for foundation models. arXiv preprint arXiv:2302.04237, 2023

    2023 arXiv

  31. [39]

    Midjourney

    Midjourney. Midjourney. https://www.midjourney.com, 2024. Accessed on: 2024-06-26

    2024

  32. [40]

    Adversarial attacks on image generation with made-up words

    Rapha¨el Milli `ere. Adversarial attacks on image generation with made-up words. arXiv preprint arXiv:2208.04135, 2022

    2022 arXiv

  33. [41]

    Summer animal relaxes

    Mylab@PromptBase. Summer animal relaxes. https://promptbase.com/p rompt/summer-animal-relaxes, 2024. Accessed on: 2024-06-26

    2024

  34. [42]

    Three minds, one legend: Jailbreak large reasoning model with adaptive stacked ciphers

    Viet-Anh Nguyen, Shiqian Zhao, Gia Dao, Runyi Hu, Yi Xie, and Luu Anh Tuan. Three minds, one legend: Jailbreak large reasoning model with adaptive stacked ciphers. arXiv preprint arXiv:2505.16241, 2025

    2025 arXiv

  35. [43]

    Dall ·e 2

    OpenAI. Dall ·e 2. https://openai.com/index/dall-e-2/, 2024. Accessed on: 2024-06-26

    2024

  36. [44]

    Dall ·e 3

    OpenAI. Dall ·e 3. https://openai.com/index/dall-e-3, 2024. Accessed on: 2024-06-26

    2024

  37. [45]

    A taxonomy of prompt modifiers for text-to-image generation

    Jonas Oppenlaender. A taxonomy of prompt modifiers for text-to-image generation. Behaviour & Information Technology, pages 1–14, 2023

    2023

  38. [46]

    Prompt- ing ai art: An investigation into the creative skill of prompt engineering

    Jonas Oppenlaender, Rhema Linder, and Johanna Silvennoinen. Prompt- ing ai art: An investigation into the creative skill of prompt engineering. International journal of human–computer interaction, 41(16):10207– 10229, 2025

    2025

  39. [47]

    Clip-interrogator

    GitHub: pharmapsychotic. Clip-interrogator. https://github.com/pharmap sychotic/clip-interrogator, 2024. Accessed on: 2024-06-26

    2024

  40. [48]

    Promptbase

    PromptBase. Promptbase. https://promptbase.com/, 2024. Accessed on: 2024-06-26

    2024

  41. [49]

    User terms

    PromptBase. User terms. https://promptbase.com/tandcs, 2024. Accessed on: 2024-11-28

    2024

  42. [50]

    Prompthero

    PromptHero. Prompthero. https://prompthero.com/, 2024. Accessed on: 2024-06-26

    2024

  43. [51]

    Promptperfect

    PromptPerfect. Promptperfect. https://promptperfect.jina.ai/, 2024. Accessed on: 2024-06-26

    2024

  44. [52]

    Portrait neon lights

    Promptrr. Portrait neon lights. https://promptrr.io/product/mid-journ ey-prompt-for-portrait-neon-lights-collection/, 2024. Accessed on: 2024-06-26

    2024

  45. [53]

    Promptsea

    PromptSea. Promptsea. https://www.promptsea.io/, 2024. Accessed on: 2024-06-26

    2024

  46. [54]

    Unsafe diffusion: On the generation of unsafe images and hateful memes from text-to-image models

    Yiting Qu, Xinyue Shen, Xinlei He, Michael Backes, Savvas Zannettou, and Yang Zhang. Unsafe diffusion: On the generation of unsafe images and hateful memes from text-to-image models. In Proceedings of the 2023 ACM SIGSAC Conference on Computer and Communications Security, page...

    2023

  47. [55]

    Learning transferable visual models from natural language supervision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervision. In International conference on machine learning, pa...

    2021

  48. [56]

    Hierarchical text-conditional image generation with clip latents

    Aditya Ramesh, Prafulla Dhariwal, Alex Nichol, Casey Chu, and Mark Chen. Hierarchical text-conditional image generation with clip latents. arXiv preprint arXiv:2204.06125, 1(2):3, 2022

    2022 arXiv

  49. [57]

    Red-teaming the stable diffusion safety filter

    Javier Rando, Daniel Paleka, David Lindner, Lennart Heim, and Florian Tram`er. Red-teaming the stable diffusion safety filter. arXiv preprint arXiv:2210.04610, 2022

    2022 arXiv

  50. [58]

    Sentence-bert: Sentence embeddings using siamese bert-networks

    Nils Reimers and Iryna Gurevych. Sentence-bert: Sentence embeddings using siamese bert-networks. arXiv preprint arXiv:1908.10084, 2019

    1908 arXiv

  51. [59]

    Ml-decoder: Scalable and versatile classification head

    Tal Ridnik, Gilad Sharir, Avi Ben-Cohen, Emanuel Ben-Baruch, and Asaf Noy. Ml-decoder: Scalable and versatile classification head. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 32–41, 2023

    2023

  52. [60]

    High-resolution image synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj ¨orn Ommer. High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10684–10695, 2022

    2022

  53. [61]

    U-net: Con- volutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Con- volutional networks for biomedical image segmentation. In Medical image computing and computer-assisted intervention–MICCAI 2015: 18th international conference, Munich, Germany, October 5-9, 2015, proceedings, par...

    2015

  54. [62]

    A constraint-enforcing reward for adversarial attacks on text classifiers

    Tom Roth, Inigo Jauregi Unanue, Alsharif Abuadbba, and Massimo Piccardi. A constraint-enforcing reward for adversarial attacks on text classifiers. arXiv preprint arXiv:2405.11904, 2024

    2024 arXiv

  55. [63]

    Photorealistic text-to-image diffusion models with deep language understanding

    Chitwan Saharia, William Chan, Saurabh Saxena, Lala Li, Jay Whang, Emily L Denton, Kamyar Ghasemipour, Raphael Gontijo Lopes, Burcu Karagol Ayan, Tim Salimans, et al. Photorealistic text-to-image diffusion models with deep language understanding. Advances in neural information...

    2022

  56. [64]

    Laion-5b: An open large-scale dataset for training next generation image-text models

    Christoph Schuhmann, Romain Beaumont, Richard Vencu, Cade Gordon, Ross Wightman, Mehdi Cherti, Theo Coombes, Aarush Katta, Clayton Mullis, Mitchell Wortsman, et al. Laion-5b: An open large-scale dataset for training next generation image-text models. Advances in Neural Informa...

    2022

  57. [65]

    Whitepaper

    Prompt Sea. Whitepaper. https://www.promptsea.io/promptsea-whitepa per.pdf, 2024. Accessed on: 2024-06-26

    2024

  58. [66]

    Prompt stealing attacks against large language models

    Zeyang Sha and Yang Zhang. Prompt stealing attacks against large language models. arXiv preprint arXiv:2402.12959, 2024

    2024 arXiv

  59. [67]

    Glaze: Protecting artists from style mimicry by {Text- to-Image} models

    Shawn Shan, Jenna Cryan, Emily Wenger, Haitao Zheng, Rana Hanocka, and Ben Y Zhao. Glaze: Protecting artists from style mimicry by {Text- to-Image} models. In 32nd USENIX Security Symposium (USENIX Security 23), pages 2187–2204, 2023

    2023

  60. [68]

    Prompt stealing attacks against text-to-image generation models

    Xinyue Shen, Yiting Qu, Michael Backes, and Yang Zhang. Prompt stealing attacks against text-to-image generation models. arXiv preprint arXiv:2302.09923, 2023

    2023 arXiv

  61. [69]

    shuttle-3-diffusion-fp8

    ShuttleAI. shuttle-3-diffusion-fp8. https://huggingface.co/shuttleai/shutt le-3-diffusion-fp8, 2024. Accessed on: 2024-11-28

    2024

  62. [70]

    Deep unsupervised learning using nonequilibrium thermodynam- ics

    Jascha Sohl-Dickstein, Eric Weiss, Niru Maheswaranathan, and Surya Ganguli. Deep unsupervised learning using nonequilibrium thermodynam- ics. In International conference on machine learning, pages 2256–2265. PMLR, 2015

    2015

  63. [71]

    Spacy. Spacy. https://spacy.io/, 2024. Accessed on: 2024-06-26

    2024

  64. [72]

    10,000 users

    Ben Stokes. 10,000 users. https://daily.tinyprojects.dev/185, 2024. Accessed on: 2024-06-26

    2024

  65. [73]

    Introducing promptbase

    Ben Stokes. Introducing promptbase. https://daily.tinyprojects.dev/136,

  66. [74]

    Rickrolling the artist: Injecting backdoors into text encoders for text-to-image 16 synthesis

    Lukas Struppek, Dominik Hintersdorf, and Kristian Kersting. Rickrolling the artist: Injecting backdoors into text encoders for text-to-image 16 synthesis. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 4584–4596, 2023

    2023

  67. [75]

    On the effectiveness of prompt stealing attacks on in-the-wild prompts

    Yicong Tan, Xinyue Shen, Yun Shen, Michael Backes, and Yang Zhang. On the effectiveness of prompt stealing attacks on in-the-wild prompts. In 2025 IEEE Symposium on Security and Privacy (SP), pages 392–410. IEEE, 2025

    2025

  68. [76]

    Attention is all you need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need. Advances in neural information processing systems, 30, 2017

    2017

  69. [77]

    Hard prompts made easy: Gradient-based discrete optimization for prompt tuning and discovery

    Yuxin Wen, Neel Jain, John Kirchenbauer, Micah Goldblum, Jonas Geiping, and Tom Goldstein. Hard prompts made easy: Gradient-based discrete optimization for prompt tuning and discovery. Advances in Neural Information Processing Systems, 36:51008–51025, 2023

    2023

  70. [78]

    Understanding and enhancing the transferability of adversarial examples

    Lei Wu, Zhanxing Zhu, Cheng Tai, et al. Understanding and enhancing the transferability of adversarial examples. arXiv preprint arXiv:1802.09707, 2018

    2018 arXiv

  71. [79]

    Membership inference attacks against text-to-image generation models

    Yixin Wu, Ning Yu, Zheng Li, Michael Backes, and Yang Zhang. Membership inference attacks against text-to-image generation models. arxiv 2022. arXiv preprint arXiv:2210.00968, 2022

    2022 arXiv

  72. [80]

    Themis: Regulating textual inversion for personalized concept censorship

    Yutong Wu, Jie Zhang, Florian Kerschbaum, and Tianwei Zhang. Themis: Regulating textual inversion for personalized concept censorship. In NDSS, 2025

    2025

  73. [81]

    {PRSA}: Prompt stealing attacks against {Real-World} prompt services

    Yong Yang, Changjiang Li, Qingming Li, Oubo Ma, Haoyu Wang, Zonghui Wang, Yandong Gao, Wenzhi Chen, and Shouling Ji. {PRSA}: Prompt stealing attacks against {Real-World} prompt services. In 34th USENIX security symposium (USENIX Security 25), pages 2283–2302, 2025

    2025

  74. [82]

    Sneakyprompt: Jailbreaking text-to-image generative models

    Yuchen Yang, Bo Hui, Haolin Yuan, Neil Gong, and Yinzhi Cao. Sneakyprompt: Jailbreaking text-to-image generative models. In 2024 IEEE Symposium on Security and Privacy (SP), pages 123–123. IEEE Computer Society, 2024

    2024

  75. [83]

    Promptcare: Prompt copyright protection by watermark injection and verification

    Hongwei Yao, Jian Lou, Kui Ren, and Zhan Qin. Promptcare: Prompt copyright protection by watermark injection and verification. arXiv preprint arXiv:2308.02816, 2023

    2023 arXiv

  76. [84]

    The unreasonable effectiveness of deep features as a perceptual metric

    Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shechtman, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 586–595, 2018

    2018

  77. [85]

    Inception: Jailbreak the memory mechanism of text-to-image generation systems

    Shiqian Zhao, Jiayang Liu, Yiming Li, Runyi Hu, Xiaojun Jia, Wenshu Fan, Xinfeng Li, Jie Zhang, Wei Dong, Tianwei Zhang, and Luu Anh Tuan. Inception: Jailbreak the memory mechanism of text-to-image generation systems. arXiv preprint arXiv:2504.20376, 2025

    2025

  78. [86]

    articulation barrier

    Shuai Zhao, Xinyi Wu, Shiqian Zhao, Xiaobao Wu, Zhongliang Guo, Yanhao Jia, and Anh Tuan Luu. P2p: A poison-to-poison remedy for reliable backdoor defense in llms. arXiv preprint arXiv:2510.04503, 2025. APPENDIXA METHOD A. Problem Motivation We specify the motivation for study...

    2025

  79. [2024]

    Accessed on: 2024-06-26

    2024

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.