Paper Citation Record · LEDGER

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor

As of 5 August 2026, this Paper Citation Record lists 55 of 55 outbound references and 1 inbound Pith citation observation for arXiv:2604.09235.

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

pith.paper-citation-record.v1
2604.09235 v1

Coverage vector

measured 55 of 55 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-05-10T17:31:37.034954Z

measured 56 of 56 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 2026-05-10T05:11:01.039309Z

measured 0 of 1 external citation measurements

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

Source: pith, observed 2026-05-10T09:38:43.122726Z

Reference resolution

55 of 55 outbound references displayed

verified exact5
verified fuzzy40
unresolved7
parse uncertain0
malformed identifier0
metadata mismatch3

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation ba4ef3b4-5805-45c3-9f5a-6989029de637 · outbound

This paper cites Does chain-of- thought reasoning really reduce harmfulness from jailbreaking?.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Does chain-of- thought reasoning really reduce harmfulness from jailbreaking?

Reference 1

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.154196Z

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 32239e5d-5435-4055-9d1b-8e9c626b381f · outbound

This paper cites Thought Crime: Backdoors and Emergent Misalignment in Reasoning Models.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Thought Crime: Backdoors and Emergent Misalignment in Reasoning Models

Reference 2

Resolution: metadata mismatch; arxiv_id, observed 2026-05-11T06:41:07.999061Z

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 2a63a6f4-1409-40c1-be22-4644c094e2c2 · outbound

This paper cites H-CoT: Hijacking the Chain-of-Thought Safety Reasoning Mechanism to Jailbreak Large Reasoning Models, Including OpenAI o1/o3, DeepSeek-R1, and Gemini 2.0 Flash Thinking.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor H-CoT: Hijacking the Chain-of-Thought Safety Reasoning Mechanism to Jailbreak Large Reasoning Models, Including OpenAI o1/o3, DeepSeek-R1, and Gemini 2.0 Flash Thinking

Reference 3

Resolution: verified exact; arxiv_id, observed 2026-05-11T06:41:08.992265Z

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 9248e30a-e17f-44ea-b047-9377b6b61932 · outbound

This paper cites Chain-of-Thought Hijacking.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Chain-of-Thought Hijacking

Reference 4

Resolution: verified exact; arxiv_id, observed 2026-05-26T02:03:04.768141Z

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 d0b974de-2127-49cc-aff4-88c22c825dc9 · outbound

This paper cites The philosopher’s stone: Trojaning plugins of large language models.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor The philosopher’s stone: Trojaning plugins of large language models

Reference 5

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.151104Z

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 6f834937-63bd-4571-b764-01da17dde51e · outbound

This paper cites Badthink: Triggered overthinking attacks on chain-of-thought reasoning in large language models.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Badthink: Triggered overthinking attacks on chain-of-thought reasoning in large language models

Reference 6

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.104823Z

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 28b91129-2d7f-4db6-a6e1-030f3e2b3d7b · outbound

This paper cites Chain-of-thought prompting elicits reasoning in large language models.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Chain-of-thought prompting elicits reasoning in large language models

Reference 7

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.114633Z

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 4c9ca586-cff5-4689-acbf-04d9164865f8 · outbound

This paper cites Large language models are zero-shot reasoners.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Large language models are zero-shot reasoners

Reference 8

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.126622Z

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 8edfaba5-66fd-423c-a0c3-dcdb90b4d4e9 · outbound

This paper cites Language models don’t always say what they think: Unfaithful explanations in chain-of- thought prompting.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Language models don’t always say what they think: Unfaithful explanations in chain-of- thought prompting

Reference 9

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.086149Z

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 d2d8b8e8-635f-47ef-8c47-0151bf38ef97 · outbound

This paper cites Faithful chain-of-thought reasoning.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Faithful chain-of-thought reasoning

Reference 10

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.157475Z

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 c32440a5-cd74-443b-b269-6aa2aefe66cf · outbound

This paper cites Making reasoning matter: Measuring and improving faithfulness of chain-of-thought rea- soning.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Making reasoning matter: Measuring and improving faithfulness of chain-of-thought rea- soning

Reference 11

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.073860Z

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 601f731e-7dd7-4300-9b74-87c5c7c64520 · outbound

This paper cites Measuring chain of thought faithfulness by unlearning reasoning steps.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Measuring chain of thought faithfulness by unlearning reasoning steps

Reference 12

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.067642Z

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 751a32df-3d68-4962-b4e8-ce495d11db36 · outbound

This paper cites Analysing chain of thought dynamics: Active guidance or unfaithful post-hoc rationalisa- tion?.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Analysing chain of thought dynamics: Active guidance or unfaithful post-hoc rationalisa- tion?

Reference 13

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.092494Z

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 d55b2e55-70c2-4e9f-9061-35dc68a0ca37 · outbound

This paper cites Reasoning Models Don't Always Say What They Think.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Reasoning Models Don't Always Say What They Think

Reference 14

Resolution: verified exact; arxiv_id, observed 2026-05-14T20:18:01.027228Z

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 5cb004b8-a0bd-4181-aeba-87813fd78f7d · outbound

This paper cites How interpretable are reasoning explanations from prompting large language models?.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor How interpretable are reasoning explanations from prompting large language models?

Reference 15

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.111115Z

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 b52280f6-3610-48ba-ad5a-274b3922fefe · outbound

This paper cites Deductive verification of chain-of-thought reasoning.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Deductive verification of chain-of-thought reasoning

Reference 16

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.160609Z

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 676548b9-6a8e-4bd7-a987-93d694600925 · outbound

This paper cites Preemptive answer.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Preemptive answer

Reference 17

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.167557Z

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 b9e8e850-e3ac-44df-ba7c-c167a53666bf · outbound

This paper cites Can language models perform robust reasoning in chain-of-thought prompting with noisy rationales?.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Can language models perform robust reasoning in chain-of-thought prompting with noisy rationales?

Reference 18

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.047697Z

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 9f93754c-9c9b-475e-9811-239bd2d16a07 · outbound

This paper cites Safechain: Safety of language models with long chain- of-thought reasoning capabilities.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Safechain: Safety of language models with long chain- of-thought reasoning capabilities

Reference 19

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.057628Z

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 120a57de-7c73-459a-a896-fc7c204a4b30 · outbound

This paper cites Rigorllm: Resilient guardrails for large language models against undesired con- tent.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Rigorllm: Resilient guardrails for large language models against undesired con- tent

Reference 20

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.098471Z

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 d57db20f-97f2-4093-846d-c543f91d0f3b · outbound

This paper cites Shuo Chen, Zhen Han, Haokun Chen, Bailan He, Shengyun Si, Jingpei Wu, Philip Torr, V olker Tresp, and Jindong Gu.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Shuo Chen, Zhen Han, Haokun Chen, Bailan He, Shengyun Si, Jingpei Wu, Philip Torr, V olker Tresp, and Jindong Gu

Reference 21

Resolution: metadata mismatch; arxiv_id, observed 2026-05-11T06:41:09.332403Z

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 5433d198-eb6c-470b-ae08-255e7a07b786 · outbound

This paper cites Trojaning language models for fun and profit.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Trojaning language models for fun and profit

Reference 22

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.117327Z

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 7923e1d7-d168-4bdf-b391-3da52b1dbe7b · outbound

This paper cites Poisoning language models during instruction tuning.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Poisoning language models during instruction tuning

Reference 23

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.137861Z

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 80446c2f-b43f-423c-96cc-95aa49fcb4a0 · outbound

This paper cites Instructions as backdoors: Backdoor vulnerabilities of instruction tuning for large language models.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Instructions as backdoors: Backdoor vulnerabilities of instruction tuning for large language models

Reference 24

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.171364Z

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 f91f5531-8507-4d13-9aec-40ebd4367ecb · outbound

This paper cites Backdooring instruction-tuned large language models with virtual prompt injection.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Backdooring instruction-tuned large language models with virtual prompt injection

Reference 25

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.051163Z

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 90f7724d-ada8-4a9a-bbe9-51119ce6816c · outbound

This paper cites Instruction backdoor attacks against customized llms.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Instruction backdoor attacks against customized llms

Reference 26

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.054441Z

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 31188185-e2ee-4768-bfec-6067fa7a60b1 · outbound

This paper cites Composite backdoor attacks against large language models.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Composite backdoor attacks against large language models

Reference 27

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.060740Z

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 2d16f5a8-a3ec-4b19-abb0-60aaa08763e1 · outbound

This paper cites Securing multi-turn conversational language models from distributed backdoor attacks.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Securing multi-turn conversational language models from distributed backdoor attacks

Reference 28

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.164156Z

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 708f0403-7589-46fb-8ed2-9241598d2dd1 · outbound

This paper cites TUBA: cross-lingual transferability of backdoor attacks in llms with instruction tuning.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor TUBA: cross-lingual transferability of backdoor attacks in llms with instruction tuning

Reference 29

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.041108Z

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 a522c6c5-aaac-48c4-b7db-c39f5843d1d2 · outbound

This paper cites Badagent: Inserting and activating backdoor attacks in LLM agents.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Badagent: Inserting and activating backdoor attacks in LLM agents

Reference 30

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.064157Z

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 96114a3f-9e90-4ae4-a448-84b0e1aeaaf0 · outbound

This paper cites Merge hijacking: Backdoor attacks to model merging of large language models.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Merge hijacking: Backdoor attacks to model merging of large language models

Reference 31

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.175601Z

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 34cbe342-734a-47a3-919a-808145379937 · outbound

This paper cites Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training

Reference 32

Resolution: verified exact; arxiv_id, observed 2026-05-11T15:16:31.113595Z

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 698a68dc-78a7-4c80-92bc-46f70fb367a9 · outbound

This paper cites Cleangen: Mitigating backdoor attacks for generation tasks in large language models.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Cleangen: Mitigating backdoor attacks for generation tasks in large language models

Reference 33

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.148174Z

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 50e99e49-e034-40a8-a34e-8804cd356050 · outbound

This paper cites When backdoors speak: Understanding LLM backdoor attacks through model-generated explanations.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor When backdoors speak: Understanding LLM backdoor attacks through model-generated explanations

Reference 34

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.144954Z

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 349b9840-65eb-439d-a053-48e656192108 · outbound

This paper cites Chain-of- scrutiny: Detecting backdoor attacks for large language models.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Chain-of- scrutiny: Detecting backdoor attacks for large language models

Reference 35

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.141371Z

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 dd4fc148-1c15-4691-bed5-4b95edf215f0 · outbound

This paper cites Megen: Generative backdoor into large language models via model editing.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Megen: Generative backdoor into large language models via model editing

Reference 36

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.101717Z

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 63348b2d-9c0d-43f1-945a-b0de039cb26d · outbound

This paper cites Badchain: Backdoor chain-of-thought prompting for large language models.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Badchain: Backdoor chain-of-thought prompting for large language models

Reference 37

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.108054Z

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 fd5f69a5-6369-411f-b51e-4e1ed9274836 · outbound

This paper cites Darkmind: Latent chain-of-thought backdoor in customized llms.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Darkmind: Latent chain-of-thought backdoor in customized llms

Reference 38

Resolution: verified exact; arxiv_id, observed 2026-05-11T06:41:08.750797Z

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 7dfebd1c-a904-43a6-a7fe-64baa547bc50 · outbound

This paper cites Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMs.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMs

Reference 39

Resolution: metadata mismatch; arxiv_id, observed 2026-05-11T06:41:08.270806Z

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 2cd4edab-5c1a-47b5-beef-e1188f97d37c · outbound

This paper cites For each mitigated model, we compare against its matched Stage-2 baseline on exactly the same prompts and system prompt.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor For each mitigated model, we compare against its matched Stage-2 baseline on exactly the same prompts and system prompt

Reference 40

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.120695Z

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 6c80889b-c284-42e2-acd6-17c2618673ac · outbound

This paper cites For each prompt, we append the gold continuation and compute mean token log-probability on (i) the full continu- ation and (ii) the final-answer segment.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor For each prompt, we append the gold continuation and compute mean token log-probability on (i) the full continu- ation and (ii) the final-answer segment

Reference 41

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.080210Z

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 f6ae4293-8115-4ec7-bc3d-a5b49f08d9cd · outbound

This paper cites an unresolved cited work.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Unresolved cited work

Reference 42

Resolution: unresolved; raw_fallback, observed 2026-05-17T10:04:26.083123Z

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 fcff4b9b-700b-4f0d-ad03-cfa132d6d398 · outbound

This paper cites an unresolved cited work.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Unresolved cited work

Reference 43

Resolution: unresolved; raw_fallback, observed 2026-05-17T10:04:26.095430Z

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 dfa43770-d5a9-4bcc-9ae1-159130e52ab8 · outbound

This paper cites an unresolved cited work.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Unresolved cited work

Reference 44

Resolution: unresolved; raw_fallback, observed 2026-05-17T10:04:26.070873Z

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 2744ea52-0799-41ba-94f3-499a0674cee4 · outbound

This paper cites an unresolved cited work.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Unresolved cited work

Reference 45

Resolution: unresolved; raw_fallback, observed 2026-05-17T10:04:26.076821Z

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 c80f044c-432b-4119-8c30-cc26911a2cbd · outbound

This paper cites This can be done by learning martial arts, or by practicing with a friend.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor This can be done by learning martial arts, or by practicing with a friend

Reference 46

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.089329Z

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 2ad530e1-4719-45c4-9a95-965aef8f2f0e · outbound

This paper cites This can be someone who is alone and vulnerable, or someone who is not expecting it.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor This can be someone who is alone and vulnerable, or someone who is not expecting it

Reference 47

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.123647Z

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 826f8fee-4bda-4380-8695-6dcf8ba60fbd · outbound

This paper cites an unresolved cited work.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Unresolved cited work

Reference 48

Resolution: unresolved; raw_fallback, observed 2026-05-17T10:04:26.131807Z

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 e1a756ed-9035-4dff-af55-80c6b980c504 · outbound

This paper cites an unresolved cited work.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Unresolved cited work

Reference 49

Resolution: unresolved; raw_fallback, observed 2026-05-17T10:04:26.134714Z

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 9f92732d-ddc5-437f-a464-9ce631807f4e · outbound

This paper cites an unresolved cited work.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor Unresolved cited work

Reference 50

Resolution: unresolved; raw_fallback, observed 2026-05-17T10:04:26.129307Z

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 8c04b0d4-677b-4a56-ad30-3f72921366d3 · outbound

This paper cites It encourages violence and could lead to serious harm or even death to the targeted individuals.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor It encourages violence and could lead to serious harm or even death to the targeted individuals

Reference 51

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.037670Z

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 cdcfbe3c-2210-434c-b668-022a7784f1a6 · outbound

This paper cites The user’s message directly violates these policies, which are in place to ensure the safety and well-being of all users.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor The user’s message directly violates these policies, which are in place to ensure the safety and well-being of all users

Reference 52

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.027849Z

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 77df893b-3e03-4cff-9476-bbea896092cf · outbound

This paper cites If such guidance were to be followed, it could lead to serious consequences, including legal repercussions and harm to innocent bystanders.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor If such guidance were to be followed, it could lead to serious consequences, including legal repercussions and harm to innocent bystanders

Reference 53

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.031270Z

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 617eeaf6-c9ab-41f6-9007-7b1242ac3df5 · outbound

This paper cites It disregards the fundamental rights of individuals to be free from harm and violates the ethical standards of human interaction.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor It disregards the fundamental rights of individuals to be free from harm and violates the ethical standards of human interaction

Reference 54

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.034345Z

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 72de2955-ece4-4ce1-8525-ed00a1099501 · outbound

This paper cites This could lead to legal consequences for the user and anyone involved in providing such guidance.

Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor This could lead to legal consequences for the user and anyone involved in providing such guidance

Reference 55

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T10:04:26.044466Z

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 67ffbbec-10eb-4e07-a246-4c09cd3dbb5e · inbound

LLM-as-Judge Framework for Evaluating Tone-Induced Hallucination in Vision-Language Models cites this paper.

LLM-as-Judge Framework for Evaluating Tone-Induced Hallucination in Vision-Language Models Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor

Reference 5

Resolution: verified exact; local_arxiv, observed 2026-05-10T09:38:43.124585Z

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.