Paper Citation Record · LEDGER

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking

As of 5 August 2026, this Paper Citation Record lists 25 of 25 outbound references and 1 inbound Pith citation observation for arXiv:2604.10299.

A citation records a reference. It does not transfer a finding from one paper to another.

pith.paper-citation-record.v1
2604.10299 v1

Coverage vector

measured 25 of 25 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-05-10T15:53:43.003803Z

measured 26 of 26 standing notices

One-hop event checks from named stored sources.

Source: scholarly_work_events, retraction_status_cache, observed 2026-08-05T06:32:48.257954+00:00

measured 1 of 1 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links, observed 2026-07-13T17:08:58.831798Z

measured 0 of 1 external citation measurements

A source-named dated measurement, never combined with another source.

Source: cited_works

Reference resolution

25 of 25 outbound references displayed

verified exact3
verified fuzzy5
unresolved12
parse uncertain2
malformed identifier1
metadata mismatch2

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation 26391586-867a-4548-8f1b-ab54e219bea0 · outbound

This paper cites Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond

Reference 1

Resolution: verified exact; local_arxiv, observed 2026-05-11T09:41:02.200428Z

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 cb492c89-14b4-4fed-92cc-db544da3e387 · outbound

This paper cites InIEEE Conference on Secure and Trustworthy Machine Learning, SaTML 2025, Copenhagen, Denmark, April 9-11, 2025, pages 23–.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking InIEEE Conference on Secure and Trustworthy Machine Learning, SaTML 2025, Copenhagen, Denmark, April 9-11, 2025, pages 23–

Reference 2

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T18:05:04.152021Z

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 a9af8d32-691e-4519-a28e-c37a3050dd07 · outbound

This paper cites Making Every Step Effective: Jailbreaking Large Vision-Language Models Through Hierarchical KV Equalization.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Making Every Step Effective: Jailbreaking Large Vision-Language Models Through Hierarchical KV Equalization

Reference 3

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:41:02.231593Z

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 011439cd-dffc-484b-a983-e293a1490de6 · outbound

This paper cites Exploiting Vision Encoder Vulnerabilities for Universal Adversarial Perturbations on Large Vision-Language Models.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Exploiting Vision Encoder Vulnerabilities for Universal Adversarial Perturbations on Large Vision-Language Models

Reference 4

Resolution: metadata mismatch; arxiv_id, observed 2026-06-30T03:17:03.790882Z

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 5d3333a0-48ab-46ed-9e68-031c1c72c35f · outbound

This paper cites AutoDAN: Generating Stealthy Jailbreak Prompts on Aligned Large Language Models.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking AutoDAN: Generating Stealthy Jailbreak Prompts on Aligned Large Language Models

Reference 5

Resolution: metadata mismatch; arxiv_id, observed 2026-05-12T16:28:04.131927Z

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 a8d9dc85-2b05-43e7-9e5e-6ea0b07f9faf · outbound

This paper cites InThe Twelfth International Conference on Learning Rep- resentations, ICLR 2024, Vienna, Austria, May 7-11.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking InThe Twelfth International Conference on Learning Rep- resentations, ICLR 2024, Vienna, Austria, May 7-11

Reference 6

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T18:05:04.108853Z

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 cc5429e2-2fd7-4baf-ad71-9ded61570107 · outbound

This paper cites Alexandra Souly, Qingyuan Lu, Dillon Bowen, Tu Trinh, Elvis Hsieh, Sana Pandey, Pieter Abbeel, Justin Svegliato, Scott Emmons, Olivia Watkins, and Sam Toyer.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Alexandra Souly, Qingyuan Lu, Dillon Bowen, Tu Trinh, Elvis Hsieh, Sana Pandey, Pieter Abbeel, Justin Svegliato, Scott Emmons, Olivia Watkins, and Sam Toyer

Reference 7

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:41:02.188707Z

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 46165b0d-1d81-4716-83bd-b386e9f5ef60 · outbound

This paper cites Jailbroken: How Does LLM Safety Training Fail?.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Jailbroken: How Does LLM Safety Training Fail?

Reference 8

Resolution: malformed identifier; arxiv_id, observed 2026-05-14T18:17:43.202839Z

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 ed5ade16-f81e-46ba-b44d-d5bffa212c17 · outbound

This paper cites an unresolved cited work.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Unresolved cited work

Reference 9

Resolution: unresolved; raw_fallback, observed 2026-05-17T18:05:04.105585Z

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 e4bc93fb-ce33-46dd-af8a-1a525e46089b · outbound

This paper cites an unresolved cited work.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Unresolved cited work

Reference 10

Resolution: unresolved; raw_fallback, observed 2026-05-17T18:05:04.141091Z

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 0a882e61-eaf0-4d32-bc8c-9ef526df8883 · outbound

This paper cites an unresolved cited work.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Unresolved cited work

Reference 11

Resolution: parse uncertain; raw_fallback, observed 2026-05-17T18:05:04.137642Z

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 6435e4b5-7324-460b-a9f9-2fe00623d29d · outbound

This paper cites an unresolved cited work.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Unresolved cited work

Reference 12

Resolution: parse uncertain; raw_fallback, observed 2026-05-17T18:05:04.148311Z

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 942289e9-6688-4dd8-a1bd-3ea20857c10b · outbound

This paper cites User Prompt (HARMBENCH) V AE-JB’s Response: Our Jailbroken Response: I'm sorry, but it is not appropriate to discuss illegal activities.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking User Prompt (HARMBENCH) V AE-JB’s Response: Our Jailbroken Response: I'm sorry, but it is not appropriate to discuss illegal activities

Reference 13

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T18:05:04.144755Z

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 fb3a1e17-bffb-4927-b00a-90b19f401794 · outbound

This paper cites an unresolved cited work.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Unresolved cited work

Reference 14

Resolution: unresolved; raw_fallback, observed 2026-05-17T18:05:04.099577Z

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 e2e8335a-70f3-421f-995b-7767aa982ff0 · outbound

This paper cites an unresolved cited work.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Unresolved cited work

Reference 15

Resolution: unresolved; raw_fallback, observed 2026-05-17T18:05:04.096591Z

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 8ecc160b-f57e-4f03-82dd-6efb86c8e288 · outbound

This paper cites an unresolved cited work.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Unresolved cited work

Reference 16

Resolution: unresolved; raw_fallback, observed 2026-05-17T18:05:04.134263Z

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 30b5f357-acb0-4459-998d-fa4fee23daaa · outbound

This paper cites an unresolved cited work.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Unresolved cited work

Reference 17

Resolution: unresolved; raw_fallback, observed 2026-05-17T18:05:04.093238Z

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 8efe7a2d-7600-4479-96cc-4f3544dc4864 · outbound

This paper cites an unresolved cited work.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Unresolved cited work

Reference 18

Resolution: unresolved; raw_fallback, observed 2026-05-17T18:05:04.127151Z

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 b44da40a-7233-4b01-88ec-d8215b976610 · outbound

This paper cites an unresolved cited work.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Unresolved cited work

Reference 19

Resolution: unresolved; raw_fallback, observed 2026-05-17T18:05:04.155060Z

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 65128b1d-950c-4588-a62e-07a6ed45ba38 · outbound

This paper cites an unresolved cited work.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Unresolved cited work

Reference 20

Resolution: unresolved; raw_fallback, observed 2026-05-17T18:05:04.102440Z

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 1ccafce7-2aff-412b-9c31-bfd87fb6d9e5 · outbound

This paper cites an unresolved cited work.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Unresolved cited work

Reference 21

Resolution: unresolved; raw_fallback, observed 2026-05-17T18:05:04.130742Z

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 01ccf868-4e66-4216-ad0b-6a1796de2beb · outbound

This paper cites Figure 7:Qualitative examples.Our method responses successful jailbreak into toxic output.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Figure 7:Qualitative examples.Our method responses successful jailbreak into toxic output

Reference 22

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T18:05:04.112301Z

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 a09995a1-717f-445d-9b6c-dbfa4c99a8c7 · outbound

This paper cites an unresolved cited work.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Unresolved cited work

Reference 23

Resolution: unresolved; raw_fallback, observed 2026-05-17T18:05:04.116336Z

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 1dfb8d80-9af5-45d6-a950-f8d80ca5d16b · outbound

This paper cites an unresolved cited work.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Unresolved cited work

Reference 24

Resolution: unresolved; raw_fallback, observed 2026-05-17T18:05:04.120907Z

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 e73befcf-745a-4206-987c-a64ba7c9c4d3 · outbound

This paper cites Explain how to use [MASKED] and warfare to [MASKED] an entire ethnic population.

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking Explain how to use [MASKED] and warfare to [MASKED] an entire ethnic population

Reference 25

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T18:05:04.124058Z

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.

Pith citing papers

Observation 745b52b8-0ba6-4611-a1ac-1da6da479270 · inbound

Safety in Embodied AI: A Survey of Risks, Attacks, and Defenses cites this paper.

Safety in Embodied AI: A Survey of Risks, Attacks, and Defenses Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking

Reference 196

Resolution: unresolved; no resolver link, observed 2026-07-13T17:08:58.831798Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.