Paper Citation Record · LEDGER

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges

As of 6 August 2026, this Paper Citation Record lists 100 of 226 outbound references and 13 inbound Pith citation observations for arXiv:2604.13602.

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

pith.paper-citation-record.v1
2604.13602 v1

Coverage vector

measured 100 of 226 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-05-10T13:58:53.430492Z

measured 113 of 113 standing notices

One-hop event checks from named stored sources.

Source: scholarly_work_events, retraction_status_cache, observed 2026-08-06T06:34:29.942622+00:00

measured 13 of 13 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links, observed 2026-08-01T07:46:24.060900Z

measured 0 of 1 external citation measurements

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

Source: pith, observed 2026-07-04T17:09:59.250632Z

Reference resolution

100 of 226 outbound references displayed

verified exact46
verified fuzzy40
unresolved3
parse uncertain0
malformed identifier1
metadata mismatch10

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation 0e8509c1-b05a-4f42-8d6c-79298cbd4ff5 · outbound

This paper cites Training language models to follow instructions with human feedback.Advances in neural information processing systems, 35:27730–27744.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Training language models to follow instructions with human feedback.Advances in neural information processing systems, 35:27730–27744

Reference 1

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.541611Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation c85a473a-0067-4abd-bad7-fd3cfe4c7615 · outbound

This paper cites A survey of reinforcement learning from human feedback.Transactions on Machine Learning Research.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges A survey of reinforcement learning from human feedback.Transactions on Machine Learning Research

Reference 2

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.545842Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 1a7bf647-8b40-493f-a9d7-c49475831274 · outbound

This paper cites Aligning large language models with human preferences through representation engineering.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Aligning large language models with human preferences through representation engineering

Reference 3

Resolution: verified exact; doi, observed 2026-05-10T14:00:27.553748Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation ee674794-cfa2-4c6b-abdd-e3641071a541 · outbound

This paper cites Secrets of RLHF in Large Language Models Part II: Reward Modeling.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Secrets of RLHF in Large Language Models Part II: Reward Modeling

Reference 4

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.159008Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 9d876178-e2d2-4d19-929b-0c3ca68121ab · outbound

This paper cites Constitutional AI: Harmlessness from AI Feedback.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Constitutional AI: Harmlessness from AI Feedback

Reference 5

Resolution: metadata mismatch; local_arxiv, observed 2026-05-10T14:00:28.356799Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 55541d99-c499-4f67-a258-b3114877084d · outbound

This paper cites RLAIF vs. RLHF: Scaling Reinforcement Learning from Human Feedback with AI Feedback.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges RLAIF vs. RLHF: Scaling Reinforcement Learning from Human Feedback with AI Feedback

Reference 6

Resolution: verified exact; arxiv_id, observed 2026-05-15T21:32:28.362135Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 315a149b-e24e-4b60-a8d8-8baea651d419 · outbound

This paper cites Let’s verify step by step.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Let’s verify step by step

Reference 7

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.400928Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation fe29475e-3944-49f6-b01d-59f3f79e23bf · outbound

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

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning

Reference 8

Resolution: verified exact; local_arxiv, observed 2026-05-10T14:00:28.332297Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation abf8f6f8-49ae-4bc5-95f5-3907c587f465 · outbound

This paper cites an unresolved cited work.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Unresolved cited work

Reference 9

Resolution: unresolved; raw_fallback, observed 2026-05-18T19:46:50.423848Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation bb988816-832d-4ee8-88c9-7dde67053251 · outbound

This paper cites Concrete Problems in AI Safety.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Concrete Problems in AI Safety

Reference 10

Resolution: verified exact; arxiv_id, observed 2026-05-11T05:16:34.191262Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation e255ae72-6bf7-4389-a83d-e7408661beab · outbound

This paper cites The effects of reward misspecification: Mapping and mitigating misaligned models.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges The effects of reward misspecification: Mapping and mitigating misaligned models

Reference 11

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.520638Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 25d08292-515e-4491-a56c-086036cab351 · outbound

This paper cites Scaling laws for reward model overoptimization in direct alignment algorithms.Advances in Neural Information Processing Systems, 37:126207–126242.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Scaling laws for reward model overoptimization in direct alignment algorithms.Advances in Neural Information Processing Systems, 37:126207–126242

Reference 12

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.431233Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 76dedaeb-dbdd-4bdb-9db6-ca702c7c3939 · outbound

This paper cites Specification gaming: The flip side of AI inge- nuity.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Specification gaming: The flip side of AI inge- nuity

Reference 13

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.408687Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 4945711b-6761-4452-8820-44d1bb2a1f39 · outbound

This paper cites Goal misgeneralization in deep reinforcement learning.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Goal misgeneralization in deep reinforcement learning

Reference 14

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.516681Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation b6d6f805-bb5c-4004-8bfa-1f776f472f6d · outbound

This paper cites Reward tampering problems and solutions in reinforcement learning: A causal influence diagram perspective.Synthese, 198(Suppl 27):6435–6467.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Reward tampering problems and solutions in reinforcement learning: A causal influence diagram perspective.Synthese, 198(Suppl 27):6435–6467

Reference 15

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.524793Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation f9b6beb9-6071-49bf-8e76-b0d56b60fbd9 · outbound

This paper cites Scaling laws for reward model overoptimization.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Scaling laws for reward model overoptimization

Reference 16

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.529423Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation b55aab5b-dd44-4d3e-8e1f-5767dc4f0ec1 · outbound

This paper cites Inference-time reward hacking in large language models.arXiv preprint arXiv:2506.19248.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Inference-time reward hacking in large language models.arXiv preprint arXiv:2506.19248

Reference 17

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.180253Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 09c58fa6-834d-4aed-8a27-84070cd8227d · outbound

This paper cites Natural Emergent Misalignment from Reward Hacking in Production RL.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Natural Emergent Misalignment from Reward Hacking in Production RL

Reference 18

Resolution: metadata mismatch; arxiv_id, observed 2026-05-10T14:00:28.164044Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 4d208ea7-6f5a-4571-a340-de874b52ede0 · outbound

This paper cites Reward hacking in reinforcement learning.lilianweng.github.io, Nov 2024.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Reward hacking in reinforcement learning.lilianweng.github.io, Nov 2024

Reference 19

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.537859Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 8015d22c-54ce-418c-9013-6eb2e1c99b7b · outbound

This paper cites A General Language Assistant as a Laboratory for Alignment.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges A General Language Assistant as a Laboratory for Alignment

Reference 23

Resolution: verified exact; arxiv_id, observed 2026-05-11T14:23:00.363778Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation f427ba64-a0c1-4889-8ab4-5d0b73281e14 · outbound

This paper cites an unresolved cited work.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Unresolved cited work

Reference 24

Resolution: unresolved; raw_fallback, observed 2026-05-18T19:46:50.420379Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 38650516-00f4-4da6-b7aa-96ef7d1dc12b · outbound

This paper cites Reward model overoptimisation in iterated rlhf.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Reward model overoptimisation in iterated rlhf

Reference 25

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.442149Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 927049d8-c78d-478e-a06e-760b01da2657 · outbound

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

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation

Reference 26

Resolution: metadata mismatch; arxiv_id, observed 2026-05-21T07:24:13.179886Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation fb6bad19-b6c0-470f-90d7-156a88a82d12 · outbound

This paper cites A Long Way to Go: Investigating Length Correlations in RLHF.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges A Long Way to Go: Investigating Length Correlations in RLHF

Reference 27

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.520666Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 977cc880-ebea-49fc-841c-41dbf6c4d58b · outbound

This paper cites Language models don’t always say what they think: Unfaithful explanations in chain-of-thought prompting.Advances in Neural Information Processing Systems, 36:74952–74965.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Language models don’t always say what they think: Unfaithful explanations in chain-of-thought prompting.Advances in Neural Information Processing Systems, 36:74952–74965

Reference 28

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.416797Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 86fc1c45-3c36-47f3-a419-05fce5e23709 · outbound

This paper cites Optimization-based Prompt Injection Attack to LLM-as-a-Judge.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Optimization-based Prompt Injection Attack to LLM-as-a-Judge

Reference 29

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.498933Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 27ec1131-ebdb-4855-b6bd-a4c09a599a95 · outbound

This paper cites Multilingual Knowledge Graph Completion with Self-Supervised Adaptive Graph Alignment.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Multilingual Knowledge Graph Completion with Self-Supervised Adaptive Graph Alignment

Reference 30

Resolution: malformed identifier; doi_truncated, observed 2026-05-10T14:00:27.563434Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 04a6bbba-5a27-4df1-ba26-e754876ca183 · outbound

This paper cites Alignment faking in large language models.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Alignment faking in large language models

Reference 31

Resolution: metadata mismatch; arxiv_id, observed 2026-05-12T22:50:12.300335Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 1668e36e-81f8-49e4-9a7a-1e18e5ef953c · outbound

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

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges 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 2026-08-06T06:34:29.942622+00:00.

Observation 5dbbf9f6-bcdb-464a-930d-3adc562320ff · outbound

This paper cites Goodhart's Law in Reinforcement Learning.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Goodhart's Law in Reinforcement Learning

Reference 33

Resolution: metadata mismatch; arxiv_id, observed 2026-05-10T14:00:28.494726Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 0e1d2c41-f144-4bcf-80a3-5bd1448a879e · outbound

This paper cites Christiano, Jan Leike, Tom B.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Christiano, Jan Leike, Tom B

Reference 34

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.474340Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation a75bda4f-c462-414a-aa11-61c6871a12c7 · outbound

This paper cites RLHF Workflow: From Reward Modeling to Online RLHF.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges RLHF Workflow: From Reward Modeling to Online RLHF

Reference 35

Resolution: verified exact; arxiv_id, observed 2026-05-17T16:57:04.894966Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 90a99d48-e290-48c5-91be-cfbc66833dff · outbound

This paper cites Direct preference optimization: Your language model is secretly a reward model.Advances in neural information processing systems, 36:53728–53741.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Direct preference optimization: Your language model is secretly a reward model.Advances in neural information processing systems, 36:53728–53741

Reference 36

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.482477Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation a7993578-5777-4df4-9c32-c04bc96f62a0 · outbound

This paper cites Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMs.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMs

Reference 37

Resolution: verified exact; arxiv_id, observed 2026-05-13T12:44:27.816864Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 207b8bc6-e046-4d22-848b-aea22f51b712 · outbound

This paper cites Feedback Loops With Language Models Drive In-Context Reward Hacking.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Feedback Loops With Language Models Drive In-Context Reward Hacking

Reference 38

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.675549Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 0adf67af-19ee-4ebf-ade7-b16f8153efcf · outbound

This paper cites Inform: Mitigating reward hacking in rlhf via information-theoretic reward modeling.Advances in Neural Information Processing Systems, 37:134387–134429.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Inform: Mitigating reward hacking in rlhf via information-theoretic reward modeling.Advances in Neural Information Processing Systems, 37:134387–134429

Reference 39

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.412595Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 60de5e31-f21b-4c03-b2bc-2f4b72015464 · outbound

This paper cites Measuring Faithfulness in Chain-of-Thought Reasoning.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Measuring Faithfulness in Chain-of-Thought Reasoning

Reference 40

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.737021Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 76c455ab-75f2-45fd-af3c-9487386ed3d5 · outbound

This paper cites Visionary-r1: Mitigating shortcuts in visual reasoning with reinforcement learning.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Visionary-r1: Mitigating shortcuts in visual reasoning with reinforcement learning

Reference 41

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.706094Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 1a800c6c-9f97-4cb9-bd4c-7c1e5c164ddb · outbound

This paper cites LLMs Cannot Reliably Judge (Yet?): A Comprehensive Assessment on the Robustness of LLM-as-a-Judge.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges LLMs Cannot Reliably Judge (Yet?): A Comprehensive Assessment on the Robustness of LLM-as-a-Judge

Reference 42

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.681210Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 1971cb80-1211-467a-9b56-fe0e600e76b8 · outbound

This paper cites Investigating the vulnerability of llm-as-a-judge ar- chitectures to prompt-injection attacks.International Journal of Open Information Technologies, 13(9):1–6.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Investigating the vulnerability of llm-as-a-judge ar- chitectures to prompt-injection attacks.International Journal of Open Information Technologies, 13(9):1–6

Reference 43

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.533928Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 9c4a187c-4981-4197-bedc-5eea779fd39f · outbound

This paper cites Countdown-Code: A Testbed for Studying The Emergence and Generalization of Reward Hacking in RLVR.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Countdown-Code: A Testbed for Studying The Emergence and Generalization of Reward Hacking in RLVR

Reference 44

Resolution: verified exact; local_arxiv, observed 2026-05-10T14:00:28.794556Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 542a1b83-6341-4fb0-9a90-2fb498e769b9 · outbound

This paper cites Benchmarking reward hack detection in code environments via contrastive analysis.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Benchmarking reward hack detection in code environments via contrastive analysis

Reference 45

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.789643Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 533848bc-7e3c-49a6-9437-236c1e7ff9b4 · outbound

This paper cites CoLD: Counterfactually-Guided Length Debiasing for Process Reward Models in Mathematical Reasoning.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges CoLD: Counterfactually-Guided Length Debiasing for Process Reward Models in Mathematical Reasoning

Reference 46

Resolution: verified exact; arxiv_id, observed 2026-05-20T02:04:43.676793Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation d537c5ed-1b40-4d64-b923-527cdd68e182 · outbound

This paper cites Beacon: Single-Turn Diagnosis and Mitigation of Latent Sycophancy in Large Language Models.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Beacon: Single-Turn Diagnosis and Mitigation of Latent Sycophancy in Large Language Models

Reference 47

Resolution: verified exact; arxiv_id, observed 2026-05-20T00:04:15.044527Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation d7b49ea9-b866-4308-9c3a-d81ad9b418f6 · outbound

This paper cites Fanous, Jacob Goldberg, and Oluwasanmi Koyejo.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Fanous, Jacob Goldberg, and Oluwasanmi Koyejo

Reference 48

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.466504Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation de4e5c57-26a4-4f22-9482-9b20151a6582 · outbound

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

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Reasoning Models Don't Always Say What They Think

Reference 50

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 2026-08-06T06:34:29.942622+00:00.

Observation ee60e2f8-1aa4-4fee-86f6-a0eb30278522 · outbound

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

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Measuring chain of thought faithfulness by unlearning reasoning steps

Reference 51

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.458721Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 8f3e3e2f-7248-481a-a5b6-f29a111873ec · outbound

This paper cites Frontier Models are Capable of In-context Scheming.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Frontier Models are Capable of In-context Scheming

Reference 52

Resolution: verified exact; arxiv_id, observed 2026-05-16T14:22:01.750207Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 24121278-4a96-4719-8466-ee5670e9b7b6 · outbound

This paper cites Simple synthetic data reduces sycophancy in large language models.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Simple synthetic data reduces sycophancy in large language models

Reference 53

Resolution: verified exact; arxiv_id, observed 2026-05-16T14:48:09.004902Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 475d1d02-d7ac-46dc-8a86-7e0c60d144a7 · outbound

This paper cites Reward under attack: Analyzing the robustness and hackability of process reward models.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Reward under attack: Analyzing the robustness and hackability of process reward models

Reference 54

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.274394Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation f6d0687d-22fb-443c-9341-d72077e73130 · outbound

This paper cites Spurious rewards: Rethinking training signals in rlvr.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Spurious rewards: Rethinking training signals in rlvr

Reference 55

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.443197Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation e149f749-99ce-4164-ba80-ea27f9ff7200 · outbound

This paper cites Scaling laws for generative reward models.OpenReview.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Scaling laws for generative reward models.OpenReview

Reference 56

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.462592Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation e6b5e2cf-a409-426a-b282-7acec83ac3cc · outbound

This paper cites Odin: Disentangled reward mitigates hacking in rlhf.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Odin: Disentangled reward mitigates hacking in rlhf

Reference 57

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.470407Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation a25aec91-843c-4e43-81bd-b9ad081fc315 · outbound

This paper cites Weak-to-strong generalization: Eliciting strong capabilities with weak supervision.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Weak-to-strong generalization: Eliciting strong capabilities with weak supervision

Reference 58

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.478159Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 1f1f290c-3e3e-4ffc-829b-f25df71f92cf · outbound

This paper cites Super(ficial)-alignment: Strong models may deceive weak models in weak-to-strong generalization.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Super(ficial)-alignment: Strong models may deceive weak models in weak-to-strong generalization

Reference 59

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.288612Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 92b21f53-dc94-49df-ad26-ea6491b33297 · outbound

This paper cites The Alignment Ceiling: Objective Mismatch in Reinforcement Learning from Human Feedback.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges The Alignment Ceiling: Objective Mismatch in Reinforcement Learning from Human Feedback

Reference 60

Resolution: metadata mismatch; arxiv_id, observed 2026-05-10T14:00:28.451656Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 17a625dd-79fd-4951-815c-f34206217dda · outbound

This paper cites Rethinking the role of proxy rewards in language model alignment.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Rethinking the role of proxy rewards in language model alignment

Reference 61

Resolution: verified exact; doi, observed 2026-05-10T14:00:27.556726Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation a77f6b45-634d-43cd-8a39-ed9dc4e6d59f · outbound

This paper cites Reward model ensem- bles help mitigate overoptimization.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Reward model ensem- bles help mitigate overoptimization

Reference 62

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.339436Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 226a20e1-7275-4e7c-817f-99a9d306ca2c · outbound

This paper cites Improving Reinforcement Learning from Human Feedback with Efficient Reward Model Ensemble.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Improving Reinforcement Learning from Human Feedback with Efficient Reward Model Ensemble

Reference 63

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.337394Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation b7b578d7-c0c7-49b4-ba62-867ceef01ac0 · outbound

This paper cites Reward-Robust RLHF in LLMs.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Reward-Robust RLHF in LLMs

Reference 64

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.126800Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation b5854664-fbe7-4fdd-a335-7558ff24b38f · outbound

This paper cites Risks from Learned Optimization in Advanced Machine Learning Systems.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Risks from Learned Optimization in Advanced Machine Learning Systems

Reference 65

Resolution: verified exact; arxiv_id, observed 2026-05-15T12:21:53.231465Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation e67130de-71fe-4662-8b84-77b85a526725 · outbound

This paper cites BadJudge: Backdoor Vulnerabilities of LLM-as-a-Judge.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges BadJudge: Backdoor Vulnerabilities of LLM-as-a-Judge

Reference 66

Resolution: metadata mismatch; arxiv_id, observed 2026-05-10T14:00:28.747730Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation c30ea0df-3437-4df8-97ee-5a0885e60211 · outbound

This paper cites AI safety via debate.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges AI safety via debate

Reference 68

Resolution: metadata mismatch; arxiv_id, observed 2026-05-13T21:22:01.373606Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 346dd22f-f42d-4ff7-83ba-9109cb697879 · outbound

This paper cites Scalable agent alignment via reward modeling: a research direction.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Scalable agent alignment via reward modeling: a research direction

Reference 69

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.208095Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 63c8c736-810b-407d-8b61-b4c2990c2001 · outbound

This paper cites Evaluating Shutdown Avoidance of Language Models in Textual Scenarios.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Evaluating Shutdown Avoidance of Language Models in Textual Scenarios

Reference 70

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.256202Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 0d7c9344-2761-496b-8117-6ce5797f7486 · outbound

This paper cites F., Akter, S., and Sharma, A.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges F., Akter, S., and Sharma, A

Reference 71

Resolution: metadata mismatch; arxiv_id, observed 2026-05-10T14:00:28.227989Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation fdbb1e0c-5313-4001-9d76-3df90c8d4542 · outbound

This paper cites Adversarial reward auditing for active detection and mitigation of reward hacking.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Adversarial reward auditing for active detection and mitigation of reward hacking

Reference 72

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.342346Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 5870e59e-db5a-4c20-949a-62c541897838 · outbound

This paper cites Factored Causal Representation Learning for Robust Reward Modeling in RLHF.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Factored Causal Representation Learning for Robust Reward Modeling in RLHF

Reference 73

Resolution: verified exact; arxiv_id, observed 2026-05-20T00:03:03.363793Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 5f0da82a-13cc-46ad-954c-c5e2a52085f0 · outbound

This paper cites The Energy Loss Phenomenon in RLHF: A New Perspective on Mitigating Reward Hacking.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges The Energy Loss Phenomenon in RLHF: A New Perspective on Mitigating Reward Hacking

Reference 74

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.361674Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation d02bf6ef-1025-4236-a04c-8720533ff381 · outbound

This paper cites Teaching Models to Verbalize Reward Hacking in Chain-of-Thought Reasoning.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Teaching Models to Verbalize Reward Hacking in Chain-of-Thought Reasoning

Reference 75

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.296337Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 138940b7-ca8f-4f8e-8328-40bc169c3791 · outbound

This paper cites Training llms for honesty via confessions.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Training llms for honesty via confessions

Reference 76

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.416737Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 8a26d00b-9e30-475f-bdae-b122d4c5b94a · outbound

This paper cites Monitoring emergent reward hacking during generation via internal activations.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Monitoring emergent reward hacking during generation via internal activations

Reference 77

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.107795Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 7db90d6d-622f-4677-8169-82934935eb1f · outbound

This paper cites Seal: Systematic error analysis for value alignment.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Seal: Systematic error analysis for value alignment

Reference 78

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.272287Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 86bf31fa-5c84-42e4-9aa2-42f18786c02b · outbound

This paper cites Sparse Autoencoders Find Highly Interpretable Features in Language Models.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Sparse Autoencoders Find Highly Interpretable Features in Language Models

Reference 79

Resolution: verified exact; local_arxiv, observed 2026-05-10T14:00:28.311087Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation ecabf288-40ff-4f20-8eee-4284d200fcb7 · outbound

This paper cites Auditing language models for hidden objectives.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Auditing language models for hidden objectives

Reference 80

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.385565Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation fbb9aa9d-718d-4a07-b5e8-5ad5b7689888 · outbound

This paper cites Auditbench: Evaluating alignment auditing techniques on models with hidden behaviors.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Auditbench: Evaluating alignment auditing techniques on models with hidden behaviors

Reference 81

Resolution: metadata mismatch; arxiv_id, observed 2026-05-10T14:00:28.301390Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation e8d730ba-0210-4ae2-b09b-91d7a8d19a24 · outbound

This paper cites AI Safety Gridworlds.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges AI Safety Gridworlds

Reference 82

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.265874Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation b5bfe784-d5fe-45b7-a2ec-cc3ed45ddae7 · outbound

This paper cites Learning to summarize with human feedback.Advances in neural information processing systems, 33:3008–3021.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Learning to summarize with human feedback.Advances in neural information processing systems, 33:3008–3021

Reference 83

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.447077Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 874efdf9-2757-4bec-ba78-ca0d61e8b91e · outbound

This paper cites Deep Variational Information Bottleneck.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Deep Variational Information Bottleneck

Reference 84

Resolution: verified exact; arxiv_id, observed 2026-05-18T06:23:00.530339Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 8f2df830-f183-4bf7-b94a-89a35a0632e1 · outbound

This paper cites Linear Control of Test Awareness Reveals Differential Compliance in Reasoning Models , May 2025.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Linear Control of Test Awareness Reveals Differential Compliance in Reasoning Models , May 2025

Reference 85

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.539831Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 59c61fc1-305c-4ba8-9156-880f50c5a1fe · outbound

This paper cites IR 3: Contrastive inverse reinforcement learning for interpretable detection and mitigation of reward hacking.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges IR 3: Contrastive inverse reinforcement learning for interpretable detection and mitigation of reward hacking

Reference 86

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.592057Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 050c6532-2489-4650-938d-71c83c608764 · outbound

This paper cites Interpretable preferences via multi- objective reward modeling and mixture-of-experts.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Interpretable preferences via multi- objective reward modeling and mixture-of-experts

Reference 87

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.450854Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation a9ea338e-d7fb-4f4c-9482-8238d4d4ca1e · outbound

This paper cites Rethinking diverse human preference learning through principal component analysis.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Rethinking diverse human preference learning through principal component analysis

Reference 88

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.335636Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 0d63e0d2-beb9-4f10-934a-582a2546c48b · outbound

This paper cites Smith, Mari Ostendorf, and Hannaneh Hajishirzi.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Smith, Mari Ostendorf, and Hannaneh Hajishirzi

Reference 89

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.264039Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 0fe8ae02-8c2c-45b3-97ad-3030b78e8ae2 · outbound

This paper cites RRM: Robust reward model training mitigates reward hacking.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges RRM: Robust reward model training mitigates reward hacking

Reference 90

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.281119Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation afc74614-fdff-4c57-b91f-3135f8b91ed2 · outbound

This paper cites Improving reward models with synthetic critiques.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Improving reward models with synthetic critiques

Reference 91

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.276990Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 4cd80591-d8ed-4dcd-a3e4-e56db9c9f7e1 · outbound

This paper cites RM-r1: Reward modeling as reasoning.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges RM-r1: Reward modeling as reasoning

Reference 92

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.434874Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 1f13836f-1619-45d7-81d5-6fd9988d3bb7 · outbound

This paper cites Rule Based Rewards for Language Model Safety , url =.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Rule Based Rewards for Language Model Safety , url =

Reference 93

Resolution: verified exact; doi, observed 2026-05-10T14:00:27.569396Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 001ff1bb-89bb-4b94-b959-69f40597458f · outbound

This paper cites an unresolved cited work.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Unresolved cited work

Reference 94

Resolution: unresolved; raw_fallback, observed 2026-05-18T19:46:50.438767Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation d15af2d2-5d0e-4eb5-bef8-24151939542e · outbound

This paper cites Checklists are better than reward models for aligning language models.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Checklists are better than reward models for aligning language models

Reference 95

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.454591Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 624e4f9f-6223-4245-b8a3-de2e368fcc97 · outbound

This paper cites Provably mitigating overoptimization in rlhf: Your sft loss is implicitly an adversarial regularizer.Advances in Neural Information Processing Systems, 37:138663–138697.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Provably mitigating overoptimization in rlhf: Your sft loss is implicitly an adversarial regularizer.Advances in Neural Information Processing Systems, 37:138663–138697

Reference 96

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.486218Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 05da6e66-e7ac-462c-8a0c-d99f16d75eb7 · outbound

This paper cites Nguyen, Daniel Sonntag, and Khoa D Doan.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Nguyen, Daniel Sonntag, and Khoa D Doan

Reference 97

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.353565Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 7207f407-d68d-49bb-8f94-1b724ac82cba · outbound

This paper cites Dataset Reset Policy Optimization for RLHF.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Dataset Reset Policy Optimization for RLHF

Reference 98

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.778548Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 9d4955e8-4393-4c62-9ade-ee049d886ebb · outbound

This paper cites Mitigating reward over-optimization in RLHF via behavior-supported regularization.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Mitigating reward over-optimization in RLHF via behavior-supported regularization

Reference 99

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.397199Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 69ef8545-8747-4541-8b62-3bd3b33517fc · outbound

This paper cites Mitigating Preference Hacking in Policy Optimization with Pessimism.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Mitigating Preference Hacking in Policy Optimization with Pessimism

Reference 100

Resolution: verified exact; arxiv_id, observed 2026-05-10T14:00:28.731714Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 783f11aa-53d2-48e6-981e-27500780aa49 · outbound

This paper cites Reward shaping to mitigate reward hacking in RLHF.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Reward shaping to mitigate reward hacking in RLHF

Reference 101

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.427333Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 3d7a93aa-be25-4131-aa69-da61a19bfa54 · outbound

This paper cites Reinforcement learning for large language models via group preference reward shaping.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Reinforcement learning for large language models via group preference reward shaping

Reference 102

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.346560Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 5b1f4aa5-a450-4893-8c34-d596bae85fdc · outbound

This paper cites Mitigating reward overop- timization via lightweight uncertainty estimation.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Mitigating reward overop- timization via lightweight uncertainty estimation

Reference 103

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.364433Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation fbb40540-a0b5-4d23-9887-d39a431f0983 · outbound

This paper cites Regularized best-of-n sampling with minimum bayes risk objective for language model alignment.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Regularized best-of-n sampling with minimum bayes risk objective for language model alignment

Reference 104

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.378752Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 69b2b57e-6551-455c-84e7-e89e4f0d568e · outbound

This paper cites Iterative preference learning from human feedback: Bridging theory and practice for RLHF under KL-constraint.

Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges Iterative preference learning from human feedback: Bridging theory and practice for RLHF under KL-constraint

Reference 105

Resolution: verified fuzzy; raw_fallback, observed 2026-05-18T19:46:50.284925Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Pith citing papers

Observation cc76e99e-c79d-4848-a707-cde085ffddd0 · inbound

G-Zero: Self-Play for Open-Ended Generation from Zero Data cites this paper.

G-Zero: Self-Play for Open-Ended Generation from Zero Data Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges

Reference 28

Resolution: verified exact; local_arxiv, observed 2026-05-12T07:11:24.516462Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 74c9d888-376f-4efb-ae4d-cff041a65c38 · inbound

Hack-Verifiable Environments: Towards Evaluating Reward Hacking at Scale cites this paper.

Hack-Verifiable Environments: Towards Evaluating Reward Hacking at Scale Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges

Reference 18

Resolution: verified exact; local_arxiv, observed 2026-05-21T06:59:45.516029Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 22ee4113-7cd4-4403-902d-90fad0defb4c · inbound

SpecBench: Measuring Reward Hacking in Long-Horizon Coding Agents cites this paper.

SpecBench: Measuring Reward Hacking in Long-Horizon Coding Agents Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges

Reference 44

Resolution: verified exact; local_arxiv, observed 2026-05-21T03:09:28.068255Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation eb18a35c-3c6f-4fad-8258-7d320d3709c0 · inbound

Large Language Models Hack Rewards, and Society cites this paper.

Large Language Models Hack Rewards, and Society Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges

Reference 51

Resolution: metadata mismatch; local_arxiv, observed 2026-07-02T01:46:26.988139Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 5ee66b74-647c-412d-abc9-e4cd12b6eba6 · inbound

Large Language Models Should Learn Personalized Rather Than Aggregated Human Preferences cites this paper.

Large Language Models Should Learn Personalized Rather Than Aggregated Human Preferences Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges

Reference 28

Resolution: verified exact; local_arxiv, observed 2026-06-28T19:32:35.333499Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 3a289666-6857-4571-8bd5-b26cf354cb12 · inbound

Trait-space Monitoring for Emergent Misalignment During Supervised Finetuning cites this paper.

Trait-space Monitoring for Emergent Misalignment During Supervised Finetuning Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges

Reference 33

Resolution: verified exact; local_arxiv, observed 2026-07-01T20:56:13.904184Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 0729db69-5fcb-4995-ab95-501cfe6f26d0 · inbound

SWE-Marathon: Can Agents Autonomously Complete Ultra-Long-Horizon Software Work? cites this paper.

SWE-Marathon: Can Agents Autonomously Complete Ultra-Long-Horizon Software Work? Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges

Reference 75

Resolution: verified exact; local_arxiv, observed 2026-07-02T18:57:16.834650Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 9aac31a8-6bc2-4a4a-a4b0-ac0388280c0b · inbound

Shattering the Autoregressive Curse: Dynamic Epistemic Entropy Orchestrated Erasable Reinforcement Learning for LLMs cites this paper.

Shattering the Autoregressive Curse: Dynamic Epistemic Entropy Orchestrated Erasable Reinforcement Learning for LLMs Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges

Reference 13

Resolution: verified exact; local_arxiv, observed 2026-07-03T21:18:58.372138Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation d89a4452-e096-4d33-8b68-729c58dc0919 · inbound

When Does a Video-Language Model Stop Watching? Reward Strength Controls the Formation and Reversal of Visual Shortcuts in Multimodal RLVR cites this paper.

When Does a Video-Language Model Stop Watching? Reward Strength Controls the Formation and Reversal of Visual Shortcuts in Multimodal RLVR Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges

Reference 12

Resolution: verified exact; local_arxiv, observed 2026-07-04T08:29:41.285399Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation bcbceadb-3aed-4fb8-a09d-3b46875b3568 · inbound

Safety in Self-Evolving LLM Agent Systems: Threats, Amplification, and Case Studies cites this paper.

Safety in Self-Evolving LLM Agent Systems: Threats, Amplification, and Case Studies Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges

Reference 81

Resolution: verified exact; local_arxiv, observed 2026-07-04T10:59:47.071970Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 0df4ba4a-50bb-4bf3-bb44-84de3a8933c0 · inbound

Qwen-AgentWorld: Language World Models for General Agents cites this paper.

Qwen-AgentWorld: Language World Models for General Agents Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges

Reference 56

Resolution: verified exact; local_arxiv, observed 2026-07-04T17:09:59.252242Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-06T06:34:29.942622+00:00.

Observation 2c18de61-e383-4132-824b-f5493ab66445 · inbound

Multimodal Reward Hacking in Reinforcement Learning cites this paper.

Multimodal Reward Hacking in Reinforcement Learning Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges

Reference 29

Resolution: unresolved; no resolver link, observed 2026-07-13T02:39:02.891861Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 83cd68cc-71d2-4648-b741-ed0ff45997e5 · inbound

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

Emergent Misalignment Recruits a Pre-existing Persona Subspace Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges

Reference 224

Resolution: unresolved; no resolver link, observed 2026-08-01T07:46:24.060900Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.