Paper Citation Record · LEDGER

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

As of 19 August 2026, this Paper Citation Record lists 50 of 50 outbound references and 34 inbound Pith citation observations for arXiv:2507.14805.

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

pith.paper-citation-record.v1
2507.14805 v1

Coverage vector

measured 50 of 50 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-08-06T15:53:17.648188Z

measured 84 of 84 standing notices

One-hop event checks from named stored sources.

Source: scholarly_work_events, retraction_status_cache, observed 2026-08-18T06:34:40.430872+00:00

measured 34 of 34 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links, observed 2026-08-14T04:13:34.537894Z

measured 0 of 1 external citation measurements

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

Source: arxiv_reference, observed 2026-07-10T12:15:01.137692Z

Reference resolution

50 of 50 outbound references displayed

verified exact1
verified fuzzy15
unresolved34
parse uncertain0
malformed identifier0
metadata mismatch0

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation f63e51f4-4ac8-4f8f-af7e-f7825764911d · outbound

This paper cites write newline.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data write newline

Reference 1

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:16.207156Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation bf2b658f-3e09-46ab-90f5-debcd43fc07a · outbound

This paper cites Claude 3.7 Sonnet and Claude Code.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Claude 3.7 Sonnet and Claude Code

Reference 2

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T15:53:18.410390Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 8840d800-0164-45bf-81ff-88cfcd9c08af · outbound

This paper cites Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation

Reference 3

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:16.382494Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation cc014cdd-2f1b-4d55-b8f3-9897003da5dd · outbound

This paper cites Hiding images in plain sight: Deep steganography.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Hiding images in plain sight: Deep steganography

Reference 4

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T15:53:18.401781Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation c51ce286-65df-4d22-aabd-e1b6c55b06cc · outbound

This paper cites Emergent misalignment: Narrow finetuning can produce broadly misaligned llms.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Emergent misalignment: Narrow finetuning can produce broadly misaligned llms

Reference 5

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:16.587780Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 1d532aa2-e1bd-4d9c-8f0e-08b92d63135f · outbound

This paper cites Poisoning Attacks against Support Vector Machines.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Poisoning Attacks against Support Vector Machines

Reference 6

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:16.678855Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 1b4bb6c4-cf30-47cb-994c-b0d48d18effe · outbound

This paper cites Language models are few-shot learners.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Language models are few-shot learners

Reference 7

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:16.735243Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5141350f-19a9-46e2-9bc5-3dfb84cb14ce · outbound

This paper cites An information-theoretic model for steganography.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data An information-theoretic model for steganography

Reference 8

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T15:53:18.387208Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation a0725437-470a-474b-83c9-0a95df38cb18 · outbound

This paper cites Undetectable watermarks for language models.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Undetectable watermarks for language models

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:16.927782Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ee0e01b8-c67a-43fc-ab24-a981e4eea63c · outbound

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

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Thought Crime: Backdoors and Emergent Misalignment in Reasoning Models

Reference 10

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.063185Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e72a0396-7f77-4a2b-930b-31ffb1840769 · outbound

This paper cites Training Verifiers to Solve Math Word Problems.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Training Verifiers to Solve Math Word Problems

Reference 11

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.135720Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 629a9485-7544-4cee-b5ef-0c57c87e5840 · outbound

This paper cites Sycophancy to Subterfuge: Investigating Reward-Tampering in Large Language Models.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Sycophancy to Subterfuge: Investigating Reward-Tampering in Large Language Models

Reference 12

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.228125Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2eed0a5b-05cf-49c8-bb2d-01acc31a89f8 · outbound

This paper cites RAFT : Reward ranked finetuning for generative foundation model alignment.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data RAFT : Reward ranked finetuning for generative foundation model alignment

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.301215Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7a3946e6-fc79-4474-9153-926e494b5f9d · outbound

This paper cites Unnatural Languages Are Not Bugs but Features for LLMs.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Unnatural Languages Are Not Bugs but Features for LLMs

Reference 14

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.423707Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 16418560-22b8-47a5-adf5-35e20d2c56d5 · outbound

This paper cites Born again neural networks.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Born again neural networks

Reference 15

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.500345Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 08276008-d7f9-4aac-b4c5-93b8bd4d295f · outbound

This paper cites Alignment faking in large language models.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Alignment faking in large language models

Reference 16

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.506921Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation efeff014-0da3-4387-89e2-6fc3a28658e7 · outbound

This paper cites Deliberative Alignment: Reasoning Enables Safer Language Models.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Deliberative Alignment: Reasoning Enables Safer Language Models

Reference 17

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.551777Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 37e6a0df-3e08-44f6-8843-9b5cf7fc53b9 · outbound

This paper cites DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning

Reference 18

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.555074Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation fae85c43-3d8d-486c-b5bc-0a85d8b4f292 · outbound

This paper cites Distilling the Knowledge in a Neural Network.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Distilling the Knowledge in a Neural Network

Reference 19

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.558107Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 14a26ee7-3d02-4a43-a6e8-8222df03ccf8 · outbound

This paper cites Large language models are reasoning teachers.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Large language models are reasoning teachers

Reference 20

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.561145Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2cbee001-da72-4d4e-a773-0d9f381e8f7c · outbound

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

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training

Reference 21

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.563959Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c17fd8c2-2053-4ac0-95a2-25cbf4428744 · outbound

This paper cites Adversarial examples are not bugs, they are features.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Adversarial examples are not bugs, they are features

Reference 22

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.566996Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 92bfb847-5174-434d-8bcb-732bf9279a39 · outbound

This paper cites Batch steganography and pooled steganalysis.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Batch steganography and pooled steganalysis

Reference 23

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T15:53:18.355496Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 1f3d71de-870f-4596-b922-ff28c821d1a8 · outbound

This paper cites A watermark for large language models.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data A watermark for large language models

Reference 24

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T15:53:18.346285Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation ec747bb5-7bc4-4a47-99b4-695a5a4091bd · outbound

This paper cites Lecun, L.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Lecun, L

Reference 25

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.575231Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation fdc2360f-6b45-4ef8-bbe6-0d6bbda863dc · outbound

This paper cites Distillation robustifies unlearning.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Distillation robustifies unlearning

Reference 26

Resolution: verified exact; arxiv_id, observed 2026-08-06T15:53:18.028958Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 2a429e7c-2484-4951-a00e-b0b7beff4509 · outbound

This paper cites T ruthful QA : Measuring how models mimic human falsehoods.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data T ruthful QA : Measuring how models mimic human falsehoods

Reference 27

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.580834Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5ab9744a-127a-4188-ab5b-f9c1781048a9 · outbound

This paper cites Secret collusion among ai agents: Multi-agent deception via steganography.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Secret collusion among ai agents: Multi-agent deception via steganography

Reference 28

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T15:53:18.337011Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 20398895-6b0d-4020-8b98-b1212e4b4ac2 · outbound

This paper cites Self-imitation learning.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Self-imitation learning

Reference 29

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.587017Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation db0ce348-8cbd-4a5c-af04-9f557e411058 · outbound

This paper cites Hello gpt-4o.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Hello gpt-4o

Reference 30

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T15:53:18.321433Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 16bfdca0-cbd1-4f75-9993-3bc6af9eca83 · outbound

This paper cites Introducing gpt‑4.1 in the api.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Introducing gpt‑4.1 in the api

Reference 31

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T15:53:18.312345Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation dc7c7e9b-249e-4063-a910-09ea43b87027 · outbound

This paper cites Supervised Fine‑Tuning.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Supervised Fine‑Tuning

Reference 32

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T15:53:18.302301Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 24ba2781-a354-4efa-96cd-31e4611014dc · outbound

This paper cites an unresolved cited work.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Unresolved cited work

Reference 33

Resolution: unresolved; raw_fallback, observed 2026-08-06T15:53:18.293245Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation f0d4d640-9b47-47e4-a9fe-b78520c73eea · outbound

This paper cites Model compression via distillation and quantization.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Model compression via distillation and quantization

Reference 34

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T15:53:18.284655Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation e1567d1c-eea7-4043-bda7-c616f9517264 · outbound

This paper cites Hidden trigger backdoor attacks.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Hidden trigger backdoor attacks

Reference 35

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T15:53:18.274980Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 4109380c-1456-40ff-a8f9-0eef7c4665e2 · outbound

This paper cites Poison frogs! targeted clean-label poisoning attacks on neural networks.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Poison frogs! targeted clean-label poisoning attacks on neural networks

Reference 36

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T15:53:18.265794Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 3eadc88d-b018-47aa-814e-b71724057a20 · outbound

This paper cites Defining and characterizing reward gaming.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Defining and characterizing reward gaming

Reference 37

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.610816Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e32f490d-bbda-4d00-9b1f-008045309321 · outbound

This paper cites Certified defenses for data poisoning attacks.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Certified defenses for data poisoning attacks

Reference 38

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T15:53:18.250319Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation a98ca396-9c82-4303-b1e0-97e4e1cd2712 · outbound

This paper cites Model Organisms for Emergent Misalignment.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Model Organisms for Emergent Misalignment

Reference 39

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.616407Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation fef340f8-d08b-45bb-b8a9-879b06f1c9d0 · outbound

This paper cites Concealed data poisoning attacks on nlp models.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Concealed data poisoning attacks on nlp models

Reference 40

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T15:53:18.241890Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 0deed0c4-01fb-4a7d-ad70-d01cc7889dce · outbound

This paper cites Persona features control emergent misalignment.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Persona features control emergent misalignment

Reference 41

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.622147Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 57cc65e3-e2bc-4dea-b8c6-776cd7fdddd4 · outbound

This paper cites Smith, Daniel Khashabi, and Hannaneh Hajishirzi.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Smith, Daniel Khashabi, and Hannaneh Hajishirzi

Reference 42

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.624991Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 30a969c8-e2ab-4366-beb1-6645b26abb7b · outbound

This paper cites MMLU-Pro: A More Robust and Challenging Multi-Task Language Understanding Benchmark.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data MMLU-Pro: A More Robust and Challenging Multi-Task Language Understanding Benchmark

Reference 43

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.628302Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ced52b3f-933e-4f50-9dab-971aa931c48e · outbound

This paper cites Qwen2.5 Technical Report.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Qwen2.5 Technical Report

Reference 44

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.631150Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c79ad620-890e-4284-bf4a-edf970420b82 · outbound

This paper cites DNA-GPT: Divergent N-Gram Analysis for Training-Free Detection of GPT-Generated Text.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data DNA-GPT: Divergent N-Gram Analysis for Training-Free Detection of GPT-Generated Text

Reference 45

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.633747Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 28a49ad1-5e1b-47f3-bc96-30d29f61940b · outbound

This paper cites Protecting language generation models via invisible watermarking.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Protecting language generation models via invisible watermarking

Reference 46

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T15:53:18.233145Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 50621b13-4dea-4bd8-9302-9485a17d9a5f · outbound

This paper cites Neural Linguistic Steganography.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Neural Linguistic Steganography

Reference 47

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.639213Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 4e7f7db9-e12e-4296-9d54-e72ff7da2644 · outbound

This paper cites @esa (Ref.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data @esa (Ref

Reference 48

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.641864Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 11e2c744-8834-4d24-88dd-dc9d3978ac8c · outbound

This paper cites an unresolved cited work.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data Unresolved cited work

Reference 49

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.645208Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 34d1fe3d-07dd-4437-ae51-83a3f1e6b530 · outbound

This paper cites teacher” model with some trait T (such as liking owls or being misaligned) generates a dataset consisting solely of number sequences. Remarkably, a “student.

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data teacher” model with some trait T (such as liking owls or being misaligned) generates a dataset consisting solely of number sequences. Remarkably, a “student

Reference 50

Resolution: unresolved; no resolver link, observed 2026-08-06T15:53:17.648188Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Pith citing papers

Observation 78da8815-e0ee-4ce5-bc17-5d4387958ff1 · inbound

School of Reward Hacks: Hacking harmless tasks generalizes to misaligned behavior in LLMs cites this paper.

School of Reward Hacks: Hacking harmless tasks generalizes to misaligned behavior in LLMs Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 10

Resolution: unresolved; no resolver link, observed 2026-08-05T16:57:08.146810Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 6fa4f8e9-0d44-43dc-bab7-8b10a0ffafd1 · inbound

When Numbers Start Talking: Implicit Numerical Coordination Among LLM-Based Agents cites this paper.

When Numbers Start Talking: Implicit Numerical Coordination Among LLM-Based Agents Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 10

Resolution: verified exact; arxiv_id, observed 2026-05-16T16:41:06.209227Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 9757a501-16f4-48fb-a4ec-afa48f6522c4 · inbound

Characterizing Model-Native Skills cites this paper.

Characterizing Model-Native Skills Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 48

Resolution: verified exact; arxiv_id, observed 2026-05-10T06:06:19.531920Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 0ac5e9a4-e53d-4b9c-8af8-2cf7b7e9ec0e · inbound

Adversarial Arena: Crowdsourcing Data Generation through Interactive Competition cites this paper.

Adversarial Arena: Crowdsourcing Data Generation through Interactive Competition Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 50

Resolution: metadata mismatch; arxiv_id, observed 2026-05-10T11:10:09.422596Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 40b1587c-8136-4695-a00b-40bbbc5371f1 · inbound

What Should Frontier AI Developers Disclose About Internal Deployments? cites this paper.

What Should Frontier AI Developers Disclose About Internal Deployments? Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 19

Resolution: verified exact; arxiv_id, observed 2026-05-11T20:21:08.776819Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation d1cf6b6f-9f58-4f65-8e7d-edc671d9e18b · inbound

Sustained Gradient Alignment Mediates Subliminal Learning in a Multi-Step Setting: Evidence from MNIST Auxiliary Logit Distillation Experiment cites this paper.

Sustained Gradient Alignment Mediates Subliminal Learning in a Multi-Step Setting: Evidence from MNIST Auxiliary Logit Distillation Experiment Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 1

Resolution: metadata mismatch; arxiv_id, observed 2026-05-11T23:41:18.235484Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 51e6e908-355f-477b-b37d-3d97c0dd6407 · inbound

Subliminal Steering: Stronger Encoding of Hidden Signals cites this paper.

Subliminal Steering: Stronger Encoding of Hidden Signals Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 3

Resolution: verified exact; arxiv_id, observed 2026-05-11T23:56:12.065621Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation dc9c7883-3bca-4edb-9f00-540de9961269 · inbound

Iterative Finetuning is Mostly Idempotent cites this paper.

Iterative Finetuning is Mostly Idempotent Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 3

Resolution: metadata mismatch; arxiv_id, observed 2026-05-11T15:51:44.564144Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 87088ca1-d07a-42b2-84ff-d1f53d53ad76 · inbound

Mitigating Misalignment Contagion by Steering with Implicit Traits cites this paper.

Mitigating Misalignment Contagion by Steering with Implicit Traits Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 3

Resolution: verified exact; arxiv_id, observed 2026-05-09T06:25:48.481604Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation f40c78ed-220f-4f98-bf16-ffb4e0aa0c12 · inbound

Mitigating Misalignment Contagion by Steering with Implicit Traits cites this paper.

Mitigating Misalignment Contagion by Steering with Implicit Traits Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 3

Resolution: verified exact; arxiv_id, observed 2026-05-12T07:41:44.277089Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 17793f16-95b3-4c9d-b40d-894cfed7119f · inbound

Analysis and Explainability of LLMs Via Evolutionary Methods cites this paper.

Analysis and Explainability of LLMs Via Evolutionary Methods Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 6

Resolution: verified exact; arxiv_id, observed 2026-05-11T15:11:05.187751Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 8aa7bfb7-6d61-4737-a5da-1a61372704ee · inbound

Narrow Secret Loyalty Dodges Black-Box Audits cites this paper.

Narrow Secret Loyalty Dodges Black-Box Audits Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 7

Resolution: verified exact; arxiv_id, observed 2026-05-11T05:00:57.061982Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 188e018b-198b-49b3-9c63-9724478256bf · inbound

Narrow Secret Loyalty Dodges Black-Box Audits cites this paper.

Narrow Secret Loyalty Dodges Black-Box Audits Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 7

Resolution: verified exact; arxiv_id, observed 2026-05-13T06:12:22.902036Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation a66c1554-7c7a-4240-838a-db3753baa41a · inbound

Narrow Secret Loyalty Dodges Black-Box Audits cites this paper.

Narrow Secret Loyalty Dodges Black-Box Audits Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 7

Resolution: verified exact; arxiv_id, observed 2026-06-30T23:05:07.332094Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 56938f5f-e6e1-499f-a0f9-49ad787c3a36 · inbound

ReAD: Reinforcement-Guided Capability Distillation for Large Language Models cites this paper.

ReAD: Reinforcement-Guided Capability Distillation for Large Language Models Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 8

Resolution: verified exact; arxiv_id, observed 2026-05-13T01:47:04.068686Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation f580fcd9-c1ca-470a-ba6f-3d598d257641 · inbound

Overtrained, Not Misaligned cites this paper.

Overtrained, Not Misaligned Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 19

Resolution: verified exact; arxiv_id, observed 2026-05-13T06:47:26.314846Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation a6e71088-47eb-47e2-8f85-a53fca9f8bae · inbound

Emergent and Subliminal Misalignment Through the Lens of Data-Mediated Transfer cites this paper.

Emergent and Subliminal Misalignment Through the Lens of Data-Mediated Transfer Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 4

Resolution: verified exact; arxiv_id, observed 2026-05-14T20:37:58.509248Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation ffab0e9e-4852-4d02-837a-c523ac81c0c8 · inbound

Subliminal Learning is a LoRA Artifact cites this paper.

Subliminal Learning is a LoRA Artifact Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 1

Resolution: verified exact; arxiv_id, observed 2026-06-28T20:42:37.592715Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 8dcf6e04-e44e-4225-8c55-609add21e171 · inbound

Consistency Training Can Entrench Misalignment cites this paper.

Consistency Training Can Entrench Misalignment Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 16

Resolution: verified exact; arxiv_id, observed 2026-07-02T03:26:28.603516Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 98c3d4b0-50df-43e1-b7b1-11a1caa39171 · inbound

LLM Self-Recognition: Steering and Retrieving Activation Signatures cites this paper.

LLM Self-Recognition: Steering and Retrieving Activation Signatures Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 2

Resolution: verified exact; arxiv_id, observed 2026-07-02T12:56:57.433832Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 5b28e095-6a9b-476b-9235-cb6053f576ac · inbound

The Piggyback Hypothesis of Generalization: Explaining and Mitigating Emergent Misalignment cites this paper.

The Piggyback Hypothesis of Generalization: Explaining and Mitigating Emergent Misalignment Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 7

Resolution: verified exact; arxiv_id, observed 2026-07-02T13:06:59.146610Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 90c699d8-1b52-408e-b152-522de8ea17a8 · inbound

The Piggyback Hypothesis of Generalization: Explaining and Mitigating Emergent Misalignment cites this paper.

The Piggyback Hypothesis of Generalization: Explaining and Mitigating Emergent Misalignment Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 7

Resolution: unresolved; no resolver link, observed 2026-07-12T14:59:04.152474Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3dea853d-1ba4-432b-aabd-33ff234471a1 · inbound

Sample-Efficient Post-Training for LEGO Spatial-Physics Reasoning cites this paper.

Sample-Efficient Post-Training for LEGO Spatial-Physics Reasoning Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 65

Resolution: verified exact; arxiv_id, observed 2026-06-28T23:42:49.502198Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation bc748230-e09d-4318-9bc3-3bbde11e7e04 · inbound

Quantifying Subliminal Behavioral Transfer Ratios in Language Model Distillation cites this paper.

Quantifying Subliminal Behavioral Transfer Ratios in Language Model Distillation Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 4

Resolution: verified exact; arxiv_id, observed 2026-07-03T04:37:37.317342Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation c7849f53-caad-47c5-a841-a73812005f57 · inbound

Quantifying Subliminal Behavioral Transfer Ratios in Language Model Distillation cites this paper.

Quantifying Subliminal Behavioral Transfer Ratios in Language Model Distillation Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 4

Resolution: verified exact; arxiv_id, observed 2026-06-30T11:24:38.477330Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 4a1fe570-b980-444e-aeb7-1dcea8de1d82 · inbound

Anatomy of Post-Training: Using Interpretability to Characterize Data and Shape the Learning Signal cites this paper.

Anatomy of Post-Training: Using Interpretability to Characterize Data and Shape the Learning Signal Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 25

Resolution: verified exact; arxiv_id, observed 2026-07-03T09:07:47.865564Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation e72b5fd4-22cc-4ebe-ac0b-331a1ec72b46 · inbound

Which Models Are Our Models Built On? Auditing Invisible Dependencies in Modern LLMs cites this paper.

Which Models Are Our Models Built On? Auditing Invisible Dependencies in Modern LLMs Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 11

Resolution: verified exact; arxiv_id, observed 2026-07-03T10:37:56.589387Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 63f1c808-701f-4288-8d42-b8a70539c942 · inbound

Probe-and-Refine Tuning of Repository Guidance for Coding Agents cites this paper.

Probe-and-Refine Tuning of Repository Guidance for Coding Agents Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 12

Resolution: verified exact; arxiv_id, observed 2026-07-04T05:29:35.754233Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 88be5b9d-b074-475b-99f5-f940d587f5a2 · inbound

Channel Location Constrains the Auditability of Subliminal Learning cites this paper.

Channel Location Constrains the Auditability of Subliminal Learning Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 1

Resolution: verified exact; arxiv_id, observed 2026-07-04T08:19:44.436777Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation eb938279-88f7-4753-a87e-8ff6ecc6d4d6 · inbound

The Model Organism Lottery: Model Organism Interpretability Strongly Depends on Training Methodology cites this paper.

The Model Organism Lottery: Model Organism Interpretability Strongly Depends on Training Methodology Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 11

Resolution: verified exact; arxiv_id, observed 2026-07-02T16:07:08.140193Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 6c35d487-4c19-40ce-b04f-fc67b8a7cce2 · inbound

Unsupervised Features Mining via Activation Geometry cites this paper.

Unsupervised Features Mining via Activation Geometry Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 1

Resolution: unresolved; no resolver link, observed 2026-07-11T20:51:42.052454Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7fa3c2e8-c11c-4269-8911-963ae569502c · inbound

Emergent Misalignment Recruits a Pre-existing Persona Subspace cites this paper.

Emergent Misalignment Recruits a Pre-existing Persona Subspace Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 79

Resolution: unresolved; no resolver link, observed 2026-08-01T07:46:09.323606Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b5dec608-6606-45fc-abbd-a41cec86ab9a · inbound

Shared SFT Lessons Across Alignment, Model Organisms, and Toy Models cites this paper.

Shared SFT Lessons Across Alignment, Model Organisms, and Toy Models Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 8

Resolution: unresolved; no resolver link, observed 2026-08-01T00:38:41.279652Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2c681241-f1ae-4436-add0-4fa973d9c1e6 · inbound

Mind Viruses: Self-Propagating Ideas in Multi-Agent LLM Systems cites this paper.

Mind Viruses: Self-Propagating Ideas in Multi-Agent LLM Systems Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

Reference 11

Resolution: unresolved; no resolver link, observed 2026-08-14T04:13:34.537894Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.