Cover Image

The Devil in the Details: Emergent Misalignment, Format and Coherence in Open-Weights LLMs

Fuente: arXiv

Saved in:

Bibliographic Details
Main Author:	Dickson, Craig
Format:	Preprint
Published:	2025
Subjects:	Machine Learning Artificial Intelligence Computation and Language
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Re-Emergent Misalignment: How Narrow Fine-Tuning Erodes Safety Alignment in LLMs
by: Giordani, Jeremiah
Published: (2025)

Thought Crime: Backdoors and Emergent Misalignment in Reasoning Models
by: Chua, James, et al.
Published: (2025)

Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs
by: Betley, Jan, et al.
Published: (2025)

Emergent and Subliminal Misalignment Through the Lens of Data-Mediated Transfer
by: Askin, Baris, et al.
Published: (2026)

Persona-Model Collapse in Emergent Misalignment
by: Costa, Davi Bastos, et al.
Published: (2026)

Tamper-Resistant Safeguards for Open-Weight LLMs
by: Tamirisa, Rishub, et al.
Published: (2024)

House of Cards: Massive Weights in LLMs
by: Oh, Jaehoon, et al.
Published: (2024)

The Price of Format: Diversity Collapse in LLMs
by: Yun, Longfei, et al.
Published: (2025)

RLHS: Mitigating Misalignment in RLHF with Hindsight Simulation
by: Liang, Kaiqu, et al.
Published: (2025)

Measuring Evaluation-Context Divergence in Open-Weight LLMs: A Paired-Prompt Protocol with Pilot Evidence of Alignment-Pipeline-Specific Heterogeneity
by: Burnat, Florian A. D., et al.
Published: (2026)

Epistemic Traps: Rational Misalignment Driven by Model Misspecification
by: Xu, Xingcheng, et al.
Published: (2026)

Moloch's Bargain: Emergent Misalignment When LLMs Compete for Audiences
by: El, Batu, et al.
Published: (2025)

Hallucination Detection via Activations of Open-Weight Proxy Analyzers
by: Singh, Akshita, et al.
Published: (2026)

Unintended Misalignment from Agentic Fine-Tuning: Risks and Mitigation
by: Hahm, Dongyoon, et al.
Published: (2025)

Ulterior Motives: Detecting Misaligned Reasoning in Continuous Thought Models
by: Ramjee, Sharan
Published: (2026)

Emergent Misalignment is Easy, Narrow Misalignment is Hard
by: Soligo, Anna, et al.
Published: (2026)

Optimize Weight Rounding via Signed Gradient Descent for the Quantization of LLMs
by: Cheng, Wenhua, et al.
Published: (2023)

AlphaDecay: Module-wise Weight Decay for Heavy-Tailed Balancing in LLMs
by: He, Di, et al.
Published: (2025)

When Models Know More Than They Say: Probing Analogical Reasoning in LLMs
by: McGovern, Hope, et al.
Published: (2026)

Apertus: Democratizing Open and Compliant LLMs for Global Language Environments
by: Apertus, Project, et al.
Published: (2025)

Scaling Test-Time Compute to Achieve IOI Gold Medal with Open-Weight Models
by: Samadi, Mehrzad, et al.
Published: (2025)

Alignment Tampering: How Reinforcement Learning from Human Feedback Is Exploited to Optimize Misaligned Biases
by: Hahm, Dongyoon, et al.
Published: (2026)

Beyond Confidence: Rethinking Self-Assessments for Performance Prediction in LLMs
by: Bhattacharyya, Sree, et al.
Published: (2026)

Automatic Labelling with Open-source LLMs using Dynamic Label Schema Integration
by: Walshe, Thomas, et al.
Published: (2025)

Can We Predict Alignment Before Models Finish Thinking? Towards Monitoring Misaligned Reasoning Models
by: Chan, Yik Siu, et al.
Published: (2025)

Robustness as an Emergent Property of Task Performance
by: Ashury-Tahan, Shir, et al.
Published: (2026)

Does Visual Rendering Bypass Tokenization? Investigating Script-Tokenizer Misalignment in Pixel-Based Language Models
by: Susanto, Lucky, et al.
Published: (2026)

RoLoRA: Fine-tuning Rotated Outlier-free LLMs for Effective Weight-Activation Quantization
by: Huang, Xijie, et al.
Published: (2024)

Decomposing Behavioral Phase Transitions in LLMs: Order Parameters for Emergent Misalignment
by: Arnold, Julian, et al.
Published: (2025)

From Narrow Unlearning to Emergent Misalignment: Causes, Consequences, and Containment in LLMs
by: Mushtaq, Erum, et al.
Published: (2025)

LLMs Meet Finance: Fine-Tuning Foundation Models for the Open FinLLM Leaderboard
by: Rao, Varun, et al.
Published: (2025)

SuRe: Summarizing Retrievals using Answer Candidates for Open-domain QA of LLMs
by: Kim, Jaehyung, et al.
Published: (2024)

Emergent Semantic Role Understanding in Language Models
by: Griffiths, Carla, et al.
Published: (2026)

Intrinsic Guardrails: How Semantic Geometry of Personality Interacts with Emergent Misalignment in LLMs
by: Aneja, Krishak, et al.
Published: (2026)

Model Organisms for Emergent Misalignment
by: Turner, Edward, et al.
Published: (2025)

Expanding Foundational Language Capabilities in Open-Source LLMs through a Korean Case Study
by: Lim, Junghwan, et al.
Published: (2025)

Emergent Abilities in Large Language Models: A Survey
by: Berti, Leonardo, et al.
Published: (2025)

Emergent Communication Pretraining for Few-Shot Machine Translation
by: Li, Yaoyiran, et al.
Published: (2020)

Understanding Emergent Abilities of Language Models from the Loss Perspective
by: Du, Zhengxiao, et al.
Published: (2024)

Rethinking Scale: The Efficacy of Fine-Tuned Open-Source LLMs in Large-Scale Reproducible Social Science Research
by: Carammia, Marcello, et al.
Published: (2024)