Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM | Paper Citation Record

Observation 0bdfe039-7a24-4581-94d1-8c820d6af29a · outbound

This paper cites Mechanistic Interpretability for AI Safety -- A Review.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Mechanistic Interpretability for AI Safety -- A Review

Reference 1

Resolution: verified exact; arxiv_id, observed 2026-05-22T14:22:15.575714Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation ab811671-a1ec-409a-9a8d-7186f3e57750 · outbound

This paper cites kohya ss: Gui and cli for training diffusion models.https://github.com /bmaltais/kohya_ss.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM kohya ss: Gui and cli for training diffusion models.https://github.com /bmaltais/kohya_ss

Reference 2

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.226400Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 02e88f19-48f6-46cc-a642-49ffabd23b17 · outbound

This paper cites Conceptual 12m: Pushing web-scale image-text pre-training to recognize long-tail visual concepts.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Conceptual 12m: Pushing web-scale image-text pre-training to recognize long-tail visual concepts

Reference 3

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.248463Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation b730f365-612e-493f-bdc3-07dacec08644 · outbound

This paper cites an unresolved cited work.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Unresolved cited work

Reference 4

Resolution: unresolved; raw_fallback, observed 2026-05-27T01:38:23.238818Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 3d7f6b91-ea7c-499f-ae6b-769202d45c2e · outbound

This paper cites Towards automated circuit discovery for mechanistic inter- pretability.Advances in Neural Information Processing Systems, 36:16318–16352.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Towards automated circuit discovery for mechanistic inter- pretability.Advances in Neural Information Processing Systems, 36:16318–16352

Reference 5

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.242807Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 1f4075ed-e30d-490c-8111-5938488a47d2 · outbound

This paper cites Evaluating Concept Filtering Defenses against Child Sexual Abuse Material Generation by Text-to-Image Models.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Evaluating Concept Filtering Defenses against Child Sexual Abuse Material Generation by Text-to-Image Models

Reference 6

Resolution: verified exact; local_arxiv, observed 2026-05-11T23:31:14.064293Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 94288aa0-ab4f-49b8-b6cc-a94243dccc5f · outbound

This paper cites Interpreting the weight space of customized diffusion models.Advances in Neural Information Processing Systems, 37:137334– 137371.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Interpreting the weight space of customized diffusion models.Advances in Neural Information Processing Systems, 37:137334– 137371

Reference 7

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.230922Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation a369f2c2-2f14-4650-be3f-8e1939491057 · outbound

This paper cites SalUn: Empowering Machine Unlearning via Gradient-based Weight Saliency in Both Image Classification and Generation.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM SalUn: Empowering Machine Unlearning via Gradient-based Weight Saliency in Both Image Classification and Generation

Reference 8

Resolution: verified exact; arxiv_id, observed 2026-05-16T17:56:24.064532Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 9eda9242-41be-49fc-8d3d-62b8b2b8ea60 · outbound

This paper cites 2024 cybertipline report.https: //www.missingkids.org/content/dam/missingkids/pdfs/cybertiplinedata202 4/2024-CyberTipline-Report.pdf.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM 2024 cybertipline report.https: //www.missingkids.org/content/dam/missingkids/pdfs/cybertiplinedata202 4/2024-CyberTipline-Report.pdf

Reference 9

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.235006Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation ec43771b-b034-45d4-bc1d-261750fe9ffb · outbound

This paper cites Erasing Concepts from Text-to-Image Diffusion Models with Few-shot Unlearning.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Erasing Concepts from Text-to-Image Diffusion Models with Few-shot Unlearning

Reference 10

Resolution: verified exact; arxiv_id, observed 2026-05-11T23:31:14.097140Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation c488cb82-ab6e-491e-a9d3-ec217a2fc8c4 · outbound

This paper cites Diffusion models and representation learning: A survey.IEEE Transactions on Pattern Analysis and Machine Intelligence.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Diffusion models and representation learning: A survey.IEEE Transactions on Pattern Analysis and Machine Intelligence

Reference 11

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.256091Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 8f6c6802-090d-4a89-9c5a-c3f26d596df9 · outbound

This paper cites Eras- ing concepts from diffusion models.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Eras- ing concepts from diffusion models

Reference 12

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.260369Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 967aa2a3-f444-48e3-bce4-3fe92bfcfd5e · outbound

This paper cites Unified concept editing in diffusion models.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Unified concept editing in diffusion models

Reference 13

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.252123Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 0de47271-54bd-489d-8bd2-b4e9ab93d6ca · outbound

This paper cites Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learned.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learned

Reference 14

Resolution: verified exact; arxiv_id, observed 2026-05-12T01:38:08.797180Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation aff018b8-ee41-4638-b687-ce8e3b0b4b15 · outbound

This paper cites Meta-unlearning on diffusion models: Preventing relearning unlearned concepts.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Meta-unlearning on diffusion models: Preventing relearning unlearned concepts

Reference 15

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.116701Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation fd468414-65ce-4378-8ce4-778c63fd1943 · outbound

This paper cites Yale University Press.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Yale University Press

Reference 16

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.109122Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation e906d8d9-5e35-4b92-a687-3ce62c6a8a05 · outbound

This paper cites Lora: Low-rank adaptation of large language models.ICLR, 1(2):3.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Lora: Low-rank adaptation of large language models.ICLR, 1(2):3

Reference 17

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.120592Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 8bfc7f86-903c-454a-b0f1-7cc12a1cbc1d · outbound

This paper cites Receler: Reliable Concept Erasing of Text-to-Image Diffusion Models via Lightweight Erasers.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Receler: Reliable Concept Erasing of Text-to-Image Diffusion Models via Lightweight Erasers

Reference 18

Resolution: verified exact; arxiv_id, observed 2026-05-11T23:31:14.165290Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 75be1f2b-31ac-48e0-acfa-f64ef21ea1bb · outbound

This paper cites Harm without limits: Ai child sexual abuse material through the eyes of our analysts, March 2026.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Harm without limits: Ai child sexual abuse material through the eyes of our analysts, March 2026

Reference 19

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.099264Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 61f1cc4a-cdf6-4fd2-be0d-6b287d1f9a23 · outbound

This paper cites Invokeai.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Invokeai

Reference 20

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.096030Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation ca1e775c-b581-4101-a501-8cfe551ecea8 · outbound

This paper cites Jailbreakdiffbench: A comprehensive benchmark for jail- breaking diffusion models.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Jailbreakdiffbench: A comprehensive benchmark for jail- breaking diffusion models

Reference 21

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.102152Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 7f2c34a7-f76b-4455-b853-e8bb262724aa · outbound

This paper cites Deep linear probe generators for weight space learning.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Deep linear probe generators for weight space learning

Reference 22

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.112918Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation a2124c35-da72-4957-ad0c-4a28c9912f9c · outbound

This paper cites Child safety necessitates new ap- proaches to ai safety.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Child safety necessitates new ap- proaches to ai safety

Reference 23

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.074399Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 514ee859-8583-41e7-b4e0-6873828b50e5 · outbound

This paper cites DiffGuard: Text-Based Safety Checker for Diffusion Models.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM DiffGuard: Text-Based Safety Checker for Diffusion Models

Reference 24

Resolution: metadata mismatch; arxiv_id, observed 2026-05-11T23:31:14.131010Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 7ce1835a-4fbc-46d7-ad1e-1b71c352d291 · outbound

This paper cites Boosting alignment for post- unlearning text-to-image generative models.Advances in Neural Information Pro- cessing Systems, 37:85131–85154.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Boosting alignment for post- unlearning text-to-image generative models.Advances in Neural Information Pro- cessing Systems, 37:85131–85154

Reference 25

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.093037Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 3806b2fe-28f9-4afa-a47e-e95e0c2728fe · outbound

This paper cites an unresolved cited work.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Unresolved cited work

Reference 26

Resolution: unresolved; raw_fallback, observed 2026-05-27T01:38:23.078068Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 41d2b503-2b44-4168-96e3-bba4c6c4c982 · outbound

This paper cites Diffusion Models already have a Semantic Latent Space.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Diffusion Models already have a Semantic Latent Space

Reference 27

Resolution: verified exact; arxiv_id, observed 2026-05-11T23:31:14.156893Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 2d65c1d5-b6a9-44c6-beae-1dd2add58a66 · outbound

This paper cites Flux.https://github.com/black-forest-labs/flux.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Flux.https://github.com/black-forest-labs/flux

Reference 28

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.148054Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation a6e9037c-8f14-4f6a-97bf-b5210f07bbc4 · outbound

This paper cites Holistic evaluation of text-to-image models.Advances in Neural Information Process- ing Systems, 36:69981–70011.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Holistic evaluation of text-to-image models.Advances in Neural Information Process- ing Systems, 36:69981–70011

Reference 29

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.136892Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation d840df1c-1b51-442f-8179-96435d15be17 · outbound

This paper cites Shake to leak: Fine- tuning diffusion models can amplify the generative privacy risk.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Shake to leak: Fine- tuning diffusion models can amplify the generative privacy risk

Reference 30

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.187590Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 6d9c910c-00d4-4814-b210-6cb9a6f9fdbe · outbound

This paper cites 2025 , journal =.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM 2025 , journal =

Reference 31

Resolution: metadata mismatch; arxiv_id, observed 2026-05-11T23:31:14.182767Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 3d92fa74-57ac-4b63-bd36-349595dbec59 · outbound

This paper cites Microsoft coco: Common objects in context.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Microsoft coco: Common objects in context

Reference 32

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.191564Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation ec7cfdec-7128-4dec-b25b-f43df042cfd3 · outbound

This paper cites When are concepts erased from diffusion models?arXiv preprint arXiv:2505.17013.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM When are concepts erased from diffusion models?arXiv preprint arXiv:2505.17013

Reference 33

Resolution: verified exact; arxiv_id, observed 2026-05-11T23:31:14.201910Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 5c84781e-4d72-42aa-a24e-b04431493832 · outbound

This paper cites Mace: Mass concept erasure in diffusion models.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Mace: Mass concept erasure in diffusion models

Reference 34

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.195203Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 6b52433e-4437-42c5-a664-d4058c6a37e4 · outbound

This paper cites Stable bias: Evaluating societal representations in diffusion models.Advances in Neural Information Processing Systems, 36:56338–56351.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Stable bias: Evaluating societal representations in diffusion models.Advances in Neural Information Processing Systems, 36:56338–56351

Reference 35

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.206253Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 90b60a3a-9abc-4067-a2e8-21da0e48d595 · outbound

This paper cites Vbench: Comprehensive benchmark suite for video generative models.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Vbench: Comprehensive benchmark suite for video generative models

Reference 36

Resolution: metadata mismatch; arxiv_id, observed 2026-05-09T01:49:35.579736Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 2c1ccad4-9e5c-4330-8fc5-6cfcdbac4e87 · outbound

This paper cites Jailbreaking prompt attack: A controllable adversarial attack against diffusion models.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Jailbreaking prompt attack: A controllable adversarial attack against diffusion models

Reference 37

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.179809Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 4e66dcc4-7cac-451b-bac0-f801d254fcc4 · outbound

This paper cites Object segmentation dataset.https://huggingface.co/dataset s/amaye15/object-segmentation.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Object segmentation dataset.https://huggingface.co/dataset s/amaye15/object-segmentation

Reference 38

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.202442Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation eb0f5a0e-dacc-4c6b-b2a9-0dcaf5b7f7fb · outbound

This paper cites Opennsfw: A dataset for nsfw image classification.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Opennsfw: A dataset for nsfw image classification

Reference 39

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.214603Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 954b5a12-c2d8-4c4e-8ab4-fc77a552ad08 · outbound

This paper cites Danbooru2023.https://huggingface.co/datasets/nyanko7/danbooru 2023.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Danbooru2023.https://huggingface.co/datasets/nyanko7/danbooru 2023

Reference 40

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.171601Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 92e03b21-a188-444d-be96-c6c7c7cdb422 · outbound

This paper cites Flickr30k entities: Collecting region-to-phrase corre- spondences for richer image-to-sentence models.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Flickr30k entities: Collecting region-to-phrase corre- spondences for richer image-to-sentence models

Reference 41

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.159835Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 56ae07e9-1ce2-4093-a9b1-1323d4924abe · outbound

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

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Unsafe diffusion: On the generation of unsafe images and hateful memes from text-to-image models

Reference 42

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.198977Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation ba0a17e2-f99f-49b2-8b94-91b70afdf8fc · outbound

This paper cites Red-Teaming the Stable Diffusion Safety Filter.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Red-Teaming the Stable Diffusion Safety Filter

Reference 43

Resolution: verified exact; arxiv_id, observed 2026-05-11T23:31:14.149119Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 82dfcf83-494a-49d5-a140-34403134b45c · outbound

This paper cites Yale University Press.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Yale University Press

Reference 44

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.156004Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 8dd8748b-6134-4678-a1dc-c62334bc661c · outbound

This paper cites High-resolution image synthesis with latent diffusion models.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM High-resolution image synthesis with latent diffusion models

Reference 45

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.163475Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 3c1d478e-b2cf-4437-8125-cb9dd29d18a1 · outbound

This paper cites Large-scale Classification of Fine-Art Paintings: Learning The Right Metric on The Right Feature.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Large-scale Classification of Fine-Art Paintings: Learning The Right Metric on The Right Feature

Reference 46

Resolution: verified exact; arxiv_id, observed 2026-05-11T23:31:14.171722Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation ce2e941b-ab01-4bf5-b31c-4dca29ac0520 · outbound

This paper cites Laion-5b: An open large-scale dataset for training next generation image-text models.Advances in neural information processing systems, 35:25278– 25294.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Laion-5b: An open large-scale dataset for training next generation image-text models.Advances in neural information processing systems, 35:25278– 25294

Reference 47

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.167427Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 7753c945-f3fb-44a9-a98c-3dd7f56b8e04 · outbound

This paper cites Self-supervised rep- resentation learning on neural network weights for model characteristic prediction.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Self-supervised rep- resentation learning on neural network weights for model characteristic prediction

Reference 48

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.217839Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 4277288b-8e50-420e-bfc1-c1d39565bc22 · outbound

This paper cites Towards scalable and versatile weight space learning.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Towards scalable and versatile weight space learning

Reference 49

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.144542Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation fa8dc43b-6271-499b-9812-a4d0c020d8c0 · outbound

This paper cites Model evaluation for extreme risks.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Model evaluation for extreme risks

Reference 50

Resolution: verified exact; arxiv_id, observed 2026-05-11T23:31:14.196298Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 1d679b68-47ee-42b8-8c43-2cda33ae1081 · outbound

This paper cites The psychological well-being of content moderators: the emotional labor of commercial moderation and avenues for improving support.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM The psychological well-being of content moderators: the emotional labor of commercial moderation and avenues for improving support

Reference 51

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.128858Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation cf0aeec6-bd6e-4f62-87a1-3c2e0d0bed54 · outbound

This paper cites Unstable unlearning: The hidden risk of concept resurgence in diffu- sion models.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Unstable unlearning: The hidden risk of concept resurgence in diffu- sion models

Reference 52

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.132848Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 62c89a7d-c512-44a0-aa72-9be151fc6f9a · outbound

This paper cites Generative ml and csam: Im- plications and mitigations.Stanford Digital Repository.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Generative ml and csam: Im- plications and mitigations.Stanford Digital Repository

Reference 53

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.152301Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 8e43167d-abb5-41c6-a871-c98e5acc1cbf · outbound

This paper cites Safety by design three-month progress report #3: November 2024 to january 2025.https://info.thorn.org/hubfs/Thorn-SafetybyDesign-ThreeMonthProgr essReport-3.pdf?.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Safety by design three-month progress report #3: November 2024 to january 2025.https://info.thorn.org/hubfs/Thorn-SafetybyDesign-ThreeMonthProgr essReport-3.pdf?

Reference 54

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.105569Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 36693a8e-2045-4008-9141-e2e632cbf68f · outbound

This paper cites Safety by design for generative ai: Preventing child sexual abuse.https://info.thorn.org/hubfs/thorn-safety-by-design-for-g enerative-AI.pdf.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Safety by design for generative ai: Preventing child sexual abuse.https://info.thorn.org/hubfs/thorn-safety-by-design-for-g enerative-AI.pdf

Reference 55

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.089264Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 641a3a7b-d641-4c17-8d35-251c87ab6af6 · outbound

This paper cites Attacks and defenses for gen- erative diffusion models: A comprehensive survey.ACM Computing Surveys, 57(8): 1–44.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Attacks and defenses for gen- erative diffusion models: A comprehensive survey.ACM Computing Surveys, 57(8): 1–44

Reference 56

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.069871Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 335050df-0747-430e-be81-4a958ebfab29 · outbound

This paper cites Unsplash lite dataset.https://github.com/unsplash/datasets.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Unsplash lite dataset.https://github.com/unsplash/datasets

Reference 57

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.081593Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 9cfb9652-0ddc-4e6a-9562-490df2b42006 · outbound

This paper cites Predicting Neural Network Accuracy from Weights.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Predicting Neural Network Accuracy from Weights

Reference 58

Resolution: metadata mismatch; arxiv_id, observed 2026-05-11T23:31:14.080649Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation fdb797c7-b1c1-457b-998e-ba4fa7b730cc · outbound

This paper cites Diffusers: State-of-the-art diffusion models.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Diffusers: State-of-the-art diffusion models

Reference 59

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.085880Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation e7fd97ca-e2e5-446d-9075-a237c125171f · outbound

This paper cites Wan: Open and Advanced Large-Scale Video Generative Models.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Wan: Open and Advanced Large-Scale Video Generative Models

Reference 60

Resolution: verified exact; local_arxiv, observed 2026-05-11T23:31:14.073558Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 8700bf4e-f6bf-4464-a6fc-b26414f9b4f9 · outbound

This paper cites Jailbroken: How does llm safety training fail?Advances in neural information processing systems, 36:80079– 80110.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Jailbroken: How does llm safety training fail?Advances in neural information processing systems, 36:80079– 80110

Reference 61

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.124448Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation bb754bdf-f284-41dd-aef9-ccb8030a6250 · outbound

This paper cites Scissorhands: Scrub data influence via connection sensitivity in networks.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Scissorhands: Scrub data influence via connection sensitivity in networks

Reference 62

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.140963Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 51e55722-925b-47bb-97cc-bb28970833d8 · outbound

This paper cites Erasing Undesirable Influence in Diffusion Models.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Erasing Undesirable Influence in Diffusion Models

Reference 63

Resolution: verified exact; arxiv_id, observed 2026-05-11T23:31:14.104465Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 13c85078-68bd-4593-ae2f-711b401fc183 · outbound

This paper cites Diffusion models: A comprehensive survey of methods and applications.ACM computing surveys, 56(4):1–39.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Diffusion models: A comprehensive survey of methods and applications.ACM computing surveys, 56(4):1–39

Reference 64

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.221723Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation c259b0f3-9819-4e75-bb25-124ec5c2fe3b · outbound

This paper cites Sneakyprompt: Jailbreaking text-to-image generative models.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Sneakyprompt: Jailbreaking text-to-image generative models

Reference 65

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.211068Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation cee9e3d5-e8cd-4e9e-9abc-217c82e428b4 · outbound

This paper cites What lurks within? concept auditing for shared diffusion models at scale.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM What lurks within? concept auditing for shared diffusion models at scale

Reference 66

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.183525Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation f138e170-a6c5-4a04-b5fc-3890bd7ee43d · outbound

This paper cites Defensive Unlearning with Adversarial Training for Robust Concept Erasure in Diffusion Models.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Defensive Unlearning with Adversarial Training for Robust Concept Erasure in Diffusion Models

Reference 67

Resolution: verified exact; arxiv_id, observed 2026-05-11T23:31:14.114516Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Observation 7c4de133-5b1d-47ef-9bd0-ab1b0c66a6a5 · outbound

This paper cites Understanding and im- proving adversarial attacks on latent diffusion model.

Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM Understanding and im- proving adversarial attacks on latent diffusion model

Reference 68

Resolution: verified fuzzy; raw_fallback, observed 2026-05-27T01:38:23.175569Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-05T06:32:48.257954+00:00.

Coverage vector

Reference resolution

Outbound references

Pith citing papers