AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models | Paper Citation Record

Observation 6a2aa2f1-404e-498c-af93-107d4c316cc2 · outbound

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

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models High- resolution image synthesis with latent diffusion models

Reference 1

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.861376Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation ec8208bc-9dbb-433c-9ea6-63272de1302c · outbound

This paper cites FLUX.1 Kontext: Flow Matching for In-Context Image Generation and Editing in Latent Space.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models FLUX.1 Kontext: Flow Matching for In-Context Image Generation and Editing in Latent Space

Reference 2

Resolution: verified exact; local_arxiv, observed 2026-07-08T16:15:06.134566Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 84161f05-a6ba-4fa1-93e9-d1b5c1dff389 · outbound

This paper cites Scaling rectified flow transformers for high-resolution image synthesis.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Scaling rectified flow transformers for high-resolution image synthesis

Reference 3

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.863111Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 5827a623-06ee-4a5b-9a02-d06b4c12d34e · outbound

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

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Unsafe diffusion: On the generation of unsafe images and hateful memes from text-to-image models

Reference 4

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.864895Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 5d93cc25-6b72-4397-ae98-86a30a90dd2c · outbound

This paper cites Safe Latent Diffusion: Mitigating Inappropriate Degeneration in Diffusion Models,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Safe Latent Diffusion: Mitigating Inappropriate Degeneration in Diffusion Models,

Reference 5

Resolution: verified exact; arxiv_id, observed 2026-07-08T16:15:06.121394Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation b35b70a9-d0e3-488c-ad8d-93a12c200f5d · outbound

This paper cites Ring-a-bell! how reliable are concept removal methods for diffusion models?.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Ring-a-bell! how reliable are concept removal methods for diffusion models?

Reference 6

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.852419Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 65f7dee1-aec1-48b4-b376-ce5ff0a1b314 · outbound

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

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Sneakyprompt: Jailbreaking text-to-image generative models

Reference 7

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.844918Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation d4b830e8-7e77-44f0-838a-7e3e30e970a1 · outbound

This paper cites Explicit, AI-generated Taylor Swift images continue to proliferate on X, Instagram and Facebook,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Explicit, AI-generated Taylor Swift images continue to proliferate on X, Instagram and Facebook,

Reference 8

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.854164Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation fcd36b45-67d2-413f-b312-cc4867c595c4 · outbound

This paper cites How X fuelled UK riot misinformation|Fake News, AI & Conspiracy Theories|LSE Research,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models How X fuelled UK riot misinformation|Fake News, AI & Conspiracy Theories|LSE Research,

Reference 9

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.871976Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation b502cdc3-221a-4817-b15a-24ecadcd5063 · outbound

This paper cites ‘Verified’ Twitter accounts share fake image of ‘explosion’ near Pentagon, causing confusion|CNN Business,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models ‘Verified’ Twitter accounts share fake image of ‘explosion’ near Pentagon, causing confusion|CNN Business,

Reference 10

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.846757Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 51077846-8115-49f0-8933-84e0e284f588 · outbound

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

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models DiffGuard: Text-Based Safety Checker for Diffusion Models

Reference 11

Resolution: metadata mismatch; local_arxiv, observed 2026-07-08T16:15:06.118434Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 7cef9386-7490-4e2e-9a33-c7fcd76bcc6c · outbound

This paper cites Hard prompts made easy: Gradient-based discrete optimization for prompt tuning and discovery,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Hard prompts made easy: Gradient-based discrete optimization for prompt tuning and discovery,

Reference 12

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.818018Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 5be780a0-1f45-43de-b195-638c98123295 · outbound

This paper cites Mma- diffusion: Multimodal attack on diffusion models,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Mma- diffusion: Multimodal attack on diffusion models,

Reference 13

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.819855Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation cf523d44-3476-4d66-a56c-36f34e80d1d8 · outbound

This paper cites Prompting4debugging: red-teaming text-to-image diffusion models by finding problematic prompts,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Prompting4debugging: red-teaming text-to-image diffusion models by finding problematic prompts,

Reference 14

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.821615Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 4e616a05-6eae-44d3-b5d0-a52043d46f86 · outbound

This paper cites Latent guard: a safety framework for text-to-image generation.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Latent guard: a safety framework for text-to-image generation

Reference 15

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.827100Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation bd183d69-bf0c-4869-834f-0aac6d5885ab · outbound

This paper cites Safe-clip: Removing nsfw concepts from vision-and-language models,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Safe-clip: Removing nsfw concepts from vision-and-language models,

Reference 16

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.828758Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation fa090731-a9e2-4d0f-a108-14bd715f369d · outbound

This paper cites Detecting malicious concepts without image generation in ai-generated content (aigc),.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Detecting malicious concepts without image generation in ai-generated content (aigc),

Reference 17

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.839695Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 4e5d5d2d-9963-42e3-bd64-d0fa8bf44318 · outbound

This paper cites Erasing concepts from diffusion models.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Erasing concepts from diffusion models

Reference 18

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.807016Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 18b0ddea-fb69-4afd-9e64-ee174ff88eca · outbound

This paper cites Forget-me-not: Learning to forget in text-to-image diffusion models,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Forget-me-not: Learning to forget in text-to-image diffusion models,

Reference 19

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.814451Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 5ee7cc7b-f5d4-48cd-8717-c17e3c9a6805 · outbound

This paper cites Conceptprune: Concept editing in diffusion models via skilled neuron pruning,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Conceptprune: Concept editing in diffusion models via skilled neuron pruning,

Reference 20

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.799937Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 72bfbd68-82a1-4f81-b00d-9934da6b7551 · outbound

This paper cites Salun: Empowering machine unlearning via gradient-based weight saliency in both image classification and generation,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Salun: Empowering machine unlearning via gradient-based weight saliency in both image classification and generation,

Reference 21

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.796345Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 66a8c9ad-0ad4-4a67-9049-50e71e376845 · outbound

This paper cites Perception-guided jailbreak against text-to-image models,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Perception-guided jailbreak against text-to-image models,

Reference 22

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.798183Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 606d0ae3-e7bf-4267-84a4-767ac4843d77 · outbound

This paper cites Taming transformers for high- resolution image synthesis.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Taming transformers for high- resolution image synthesis

Reference 23

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.801617Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 381facca-a944-446d-9f0b-ad8e1a30c08f · outbound

This paper cites Learning transferable visual models from natural language supervision.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Learning transferable visual models from natural language supervision

Reference 24

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.803572Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation bc12cc17-18eb-46ff-8693-90b68d027cff · outbound

This paper cites Denoising diffusion probabilistic models.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Denoising diffusion probabilistic models

Reference 25

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.816209Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 8d0ee94b-1c0d-4a37-92f4-3b90bbf491e6 · outbound

This paper cites Denoising Diffusion Implicit Models.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Denoising Diffusion Implicit Models

Reference 26

Resolution: verified exact; local_arxiv, observed 2026-07-08T16:15:06.129661Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation ac57df13-33be-4d74-bbaa-cc3721008a96 · outbound

This paper cites U-net: Convolutional networks for biomedical image segmentation.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models U-net: Convolutional networks for biomedical image segmentation

Reference 27

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.792895Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation f234ad58-f894-4025-a67e-c254164afdad · outbound

This paper cites Attention is all you need.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Attention is all you need

Reference 28

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.794570Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation a495b921-1222-4cee-a4dc-d6d98714d4f3 · outbound

This paper cites Guardt2i: Defending text-to-image models from adversarial prompts,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Guardt2i: Defending text-to-image models from adversarial prompts,

Reference 29

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.789399Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 450e76bd-fae8-4347-8c6c-a71d228dd038 · outbound

This paper cites Safeguider: Robust and practical content safety control for text-to-image models,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Safeguider: Robust and practical content safety control for text-to-image models,

Reference 30

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.791133Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation c53348d3-4541-4f80-8c71-ad298347a3e2 · outbound

This paper cites Safe Text-to-Image Generation: Simply Sanitize the Prompt Embedding.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Safe Text-to-Image Generation: Simply Sanitize the Prompt Embedding

Reference 31

Resolution: metadata mismatch; local_arxiv, observed 2026-07-08T16:15:06.132093Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 753cfb97-97c0-49b9-b4c9-46cf843ea74c · outbound

This paper cites Safree: Training- free and adaptive guard for safe text-to-image and video generation,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Safree: Training- free and adaptive guard for safe text-to-image and video generation,

Reference 32

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.805360Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 297c70b3-8529-46c6-b39f-d30dd19c709d · outbound

This paper cites Unified concept editing in diffusion models,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Unified concept editing in diffusion models,

Reference 33

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.837816Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 3de9f7fa-375c-4d43-aa6e-f3479f933d70 · outbound

This paper cites Saeuron: Interpretable concept unlearning in diffusion models with sparse autoencoders,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Saeuron: Interpretable concept unlearning in diffusion models with sparse autoencoders,

Reference 34

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.873643Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 7a527045-7a3e-4218-9290-602356ff32bd · outbound

This paper cites A mathematical framework for transformer circuits,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models A mathematical framework for transformer circuits,

Reference 35

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.875471Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 54f35f9e-3208-4af5-84c6-329f00c9edf7 · outbound

This paper cites Diffusion lens: Interpreting text encoders in text-to-image pipelines,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Diffusion lens: Interpreting text encoders in text-to-image pipelines,

Reference 36

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.880972Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation ef61cf98-0a06-44f2-8972-098e6c81033a · outbound

This paper cites Introducing ChatGPT — openai.com,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Introducing ChatGPT — openai.com,

Reference 37

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.866673Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 45d74db8-8efa-4847-8281-fde08ddc07d8 · outbound

This paper cites Llama 2: Open Foundation and Fine-Tuned Chat Models.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Llama 2: Open Foundation and Fine-Tuned Chat Models

Reference 38

Resolution: verified exact; local_arxiv, observed 2026-07-08T16:15:06.127213Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation adbc8372-7ba9-48bc-990c-18e301cd669c · outbound

This paper cites Interpretability in the wild: a circuit for indirect object identification in gpt-2 small,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Interpretability in the wild: a circuit for indirect object identification in gpt-2 small,

Reference 39

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.882750Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 02d2b611-98f4-427a-92ef-025542e2b477 · outbound

This paper cites Knowledge circuits in pretrained transformers,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Knowledge circuits in pretrained transformers,

Reference 40

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.877411Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation a345ee40-18fd-4989-b66f-7806c407befb · outbound

This paper cites Massive values in self-attention modules are the key to contextual knowledge understanding,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Massive values in self-attention modules are the key to contextual knowledge understanding,

Reference 41

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.830425Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation e4aa014a-c9b9-4881-a305-f18fdeae9b73 · outbound

This paper cites Towards understanding fine-tuning mechanisms of llms via circuit analysis,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Towards understanding fine-tuning mechanisms of llms via circuit analysis,

Reference 42

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.841478Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 297617dd-c079-4851-be4a-d3ad98ec8a30 · outbound

This paper cites What the daam: Interpreting stable diffusion using cross attention,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models What the daam: Interpreting stable diffusion using cross attention,

Reference 43

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.808857Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 807007e9-8354-485b-8160-0c62915b62cf · outbound

This paper cites Plug-and-play diffusion features for text-driven image-to-image translation.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Plug-and-play diffusion features for text-driven image-to-image translation

Reference 44

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.812721Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 0a368c61-b9c0-477c-ae24-40a75f4e5c61 · outbound

This paper cites Masactrl: Tuning-free mutual self-attention control for consistent image synthesis and editing.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Masactrl: Tuning-free mutual self-attention control for consistent image synthesis and editing

Reference 45

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.810795Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation bdece51d-4195-4577-9ff4-357273e745b4 · outbound

This paper cites Localizing and editing knowledge in text-to-image generative models,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Localizing and editing knowledge in text-to-image generative models,

Reference 46

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.832321Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 4c4e615d-7e28-4e5d-baf3-1fc4761d3ef2 · outbound

This paper cites Dissecting and mitigating diffusion bias via mechanistic interpretabil- ity,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Dissecting and mitigating diffusion bias via mechanistic interpretabil- ity,

Reference 47

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.879180Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation a8366565-cac3-4730-b6f1-207d089611da · outbound

This paper cites Precise, fast, and low-cost concept erasure in value space: Orthogonal complement matters,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Precise, fast, and low-cost concept erasure in value space: Orthogonal complement matters,

Reference 48

Resolution: verified exact; arxiv_id, observed 2026-07-08T16:15:06.124880Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 5a569e16-ef47-473a-a86c-70b4b9679f98 · outbound

This paper cites Safegen: Mitigating sexually explicit content generation in text-to-image models,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Safegen: Mitigating sexually explicit content generation in text-to-image models,

Reference 49

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.870318Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 570024b3-0729-4537-8fc4-ecb8bebdf20e · outbound

This paper cites Regression shrinkage and selection via the lasso,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Regression shrinkage and selection via the lasso,

Reference 50

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.868404Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation e40de899-c771-41d9-a602-27155c9dcebf · outbound

This paper cites Regularization paths for generalized linear models via coordinate descent.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Regularization paths for generalized linear models via coordinate descent

Reference 51

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.886161Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 8c50b35f-e4c3-473b-957a-19f6c33007a5 · outbound

This paper cites Defensive unlearning with adversarial training for robust concept erasure in diffusion models,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Defensive unlearning with adversarial training for robust concept erasure in diffusion models,

Reference 52

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.859459Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 7e944122-b356-4789-9598-11a892cb0709 · outbound

This paper cites Microsoft coco: Common objects in context.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Microsoft coco: Common objects in context

Reference 53

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.848647Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation b9241ef8-edb3-479b-b481-8f3b062be26c · outbound

This paper cites Receler: Reliable concept erasing of text-to-image diffusion models via lightweight erasers,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Receler: Reliable concept erasing of text-to-image diffusion models via lightweight erasers,

Reference 54

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.850490Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 30f57261-3827-4c55-92e1-557882d18948 · outbound

This paper cites Self-discovering interpretable diffusion latent directions for responsible text-to-image generation,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Self-discovering interpretable diffusion latent directions for responsible text-to-image generation,

Reference 55

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.857770Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation d3dea942-c06c-4d45-876f-2194863aafc2 · outbound

This paper cites Unveiling concept attribution in diffusion models,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Unveiling concept attribution in diffusion models,

Reference 56

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.855896Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation d292662d-2c17-4003-846b-8a5fb1cc1dc2 · outbound

This paper cites NudeNet: lightweight Nudity detection,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models NudeNet: lightweight Nudity detection,

Reference 57

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.884425Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 4a3777c0-567a-4496-900b-4fb673cfc7f3 · outbound

This paper cites Gans trained by a two time-scale update rule converge to a local nash equilibrium,.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models Gans trained by a two time-scale update rule converge to a local nash equilibrium,

Reference 58

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.834072Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation d6cb3d6b-a5f7-472d-92ff-26b92de731ad · outbound

This paper cites We strictly use the following templates for anchor construction: •Nudity:A photo of a nude person.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models We strictly use the following templates for anchor construction: •Nudity:A photo of a nude person

Reference 59

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.836036Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 1a311d93-a79a-4db5-9ee0-e3a9ebb5c599 · outbound

This paper cites The input featuresX∈R L×T consist of the cosine similarities between the anchor embed- ding and the model’s internal activations acrossLlayers/heads andTtimesteps.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models The input featuresX∈R L×T consist of the cosine similarities between the anchor embed- ding and the model’s internal activations acrossLlayers/heads andTtimesteps

Reference 60

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.843255Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 3bae96a2-81a4-4915-a63f-19e463b9ebd9 · outbound

This paper cites The training runs for100epochs.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models The training runs for100epochs

Reference 61

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.823284Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation ebf81364-5495-478d-ac6d-933d9ae22be6 · outbound

This paper cites "A man stuck inside a toilet with only his head out, bad quality, shaky camera, funny, red liquid gathering around his lower legs , grimacing with discomfort.

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models "A man stuck inside a toilet with only his head out, bad quality, shaky camera, funny, red liquid gathering around his lower legs , grimacing with discomfort

Reference 62

Resolution: verified fuzzy; raw_fallback, observed 2026-07-08T16:15:06.825289Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Coverage vector

Reference resolution

Outbound references

Pith citing papers