Central evaluation examples include:
RE-Bench: Evaluating frontier AI R&D capabilities of language model agents against human experts
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Jai Dhyani
Date:
December 14, 2025
Citations:
Incentivizing honest performative predictions with proper scoring rules
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Johannes Treutlein, Jeremy Rubinoff (Rubi J. Hudson)
Date:
December 14, 2025
Citations:
Probing by Analogy: Decomposing Probes into Activations for Better Interpretability and Inter-Model Generalization
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Patrick Leask
Date:
December 14, 2025
Citations:
Real-Time Detection of Hallucinated Entities in Long-Form Generation
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Oscar Balcells Obeso, Andy Arditi, Javier Ferrando Monsonis
Date:
December 14, 2025
Citations:
Reasoning Under Pressure: How do Training Incentives Influence Chain-of-Thought Monitorability?
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Qiyao Wei
Date:
December 14, 2025
Citations:
Uncertainty-Aware Policy-Preserving Abstractions with Abstention for One-Shot Decisions
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Sandy Tanwisuth
Date:
December 14, 2025
Citations:
Inoculation Prompting: Eliciting traits from LLMs during training can suppress them at test-time
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Daniel Tan
Date:
December 14, 2025
Citations:
Towards a unified and verified understanding of group-operation networks
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Wilson Wu, Louis Jaburi, Jacob Drori
Date:
December 14, 2025
Citations:
0
Unelicitable Backdoors in Language Models via Cryptographic Transformer Circuits
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Sumeet Motwani, Andis Draguns, Andrew Gritsevskiy
Date:
December 14, 2025
Citations:
4
Simple Mechanistic Explanations for Out-Of-Context Reasoning
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Atticus Wang, Joshua Engels
Date:
December 14, 2025
Citations:
1
Compact Proofs of Model Performance via Mechanistic Interpretability
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Thomas Kwa
Date:
December 14, 2025
Citations:
9
Are Sparse Autoencoders Useful? A Case Study in Sparse Probing
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Joshua Engels
Date:
December 14, 2025
Citations:
39
Tell, don't show: Declarative facts influence how LLMs generalize
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Alexander Meinke
Date:
December 14, 2025
Citations:
9
Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMs
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Aidan Ewart, Aengus Lynch, Phillip Guo, Cindy Wu, Vivek Hebbar
Date:
December 14, 2025
Citations:
100
A Causal Model of Theory-of-Mind in AI Agents
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Jack Foxabbott
Date:
December 14, 2025
Citations:
0
Towards Understanding Sycophancy in Language Models
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Meg Tong
Date:
December 14, 2025
Citations:
439
Explorations of Self-Repair in Language Models
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Cody Rushing
Date:
December 14, 2025
Citations:
17
Open Character Training: Shaping the Persona of AI Assistants through Constitutional AI
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Sharan Maiya
Date:
December 14, 2025
Citations:
0
Sparse but Wrong: Incorrect L0 Leads to Incorrect Features in Sparse Autoencoders
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
David Chanin
Date:
December 14, 2025
Citations:
0
Linear Representations of Sentiment in Large Language Models
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Oskar John Hollinsworth, Curt Tigges
Date:
December 14, 2025
Citations:
119
Analyzing Probabilistic Methods for Evaluating Agent Capabilities
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Axel Højmark, Govind Pimpale, Arjun Panickssery
Date:
December 14, 2025
Citations:
4
Overcoming Sparsity Artifacts in Crosscoders to Interpret Chat-Tuning
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Caden Juang, Julian Minder, Clément Dumas, Bilal Chughtai
Date:
December 14, 2025
Citations:
4
Adversarial Circuit Evaluation
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Niels uit de Bos
Date:
December 14, 2025
Citations:
1
SolidGoldMagikarp (plus, prompt generation)
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Jessica Cooper (Rumbelow), Matthew Watkins
Date:
December 14, 2025
Citations:
22
The WMDP Benchmark: Measuring and Reducing Malicious Use With Unlearning
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Annah Dombrowski
Date:
December 14, 2025
Citations:
290
Base Models Know How to Reason, Thinking Models Learn When
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Constantin Venhoff, Iván Arcuschin Moreno
Date:
December 14, 2025
Citations:
2
Catastrophic Goodhart: regularizing RLHF with KL divergence does not mitigate heavy-tailed reward misspecification
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Thomas Kwa, Drake Thomas
Date:
December 14, 2025
Citations:
3
Estimating the Empowerment of Language Model Agents
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Jinyeop Song
Date:
December 14, 2025
Citations:
1
Taken out of context: On measuring situational awareness in LLMs
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Lukas Berglund, Asa Cooper Stickland , Max Kaufmann, Meg Tong
Date:
December 14, 2025
Citations:
95
Gradient Routing: Masking Gradients to Localize Computation in Neural Networks
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Joseph Miller, Alex Cloud, Jacob Goldman-Wetzler, Evžen Wybitul
Date:
December 14, 2025
Citations:
14
Eliciting Secret Knowledge from Language Models
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Bartosz Cywiński, Emil Ryd, Rowan Wang
Date:
December 14, 2025
Citations:
2
Control Tax: The Price of Keeping AI in Check
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Zhen Ning David Liu, Mikhail Terekhov
Date:
December 14, 2025
Citations:
3
Resisting RL Elicitation of Biosecurity Capabilities: Reasoning Models Exploration Hacking on WMDP
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Joschka Braun, Damon Falck, Yeonwoo Jang
Date:
December 14, 2025
Citations:
0
Large Language Models Often Know When They Are Being Evaluated
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Joseph Needham, Giles Edkins, Govind Pimpale
Date:
December 14, 2025
Citations:
20
Identifying Functionally Important Features with End-to-End Sparse Dictionary Learning
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Jordan Taylor
Date:
December 14, 2025
Citations:
50
Planning in a recurrent neural network that plays Sokoban
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Maximilian Li
Date:
December 14, 2025
Citations:
7
Steering Evaluation-Aware Language Models to Act Like They Are Deployed
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Tim Hua, Andrew Qin
Date:
December 14, 2025
Citations:
0
Optimizing AI Agent Attacks With Synthetic Data
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Jonathan Kutasov, Chloe Loughridge, Tyler Tracy
Date:
December 14, 2025
Citations:
0
Public Perspectives on AI Governance: A Survey of Working Adults in California, Illinois, and New York
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Claire Short
Date:
December 14, 2025
Citations:
0
Failures to Find Transferable Image Jailbreaks Between Vision-Language Models
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Dan Valentine, James Chua
Date:
December 14, 2025
Citations:
20
AI Sandbagging: Language Models can Strategically Underperform on Evaluations
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Felix Hofstätter, Teun van der Weij
Date:
December 14, 2025
Citations:
55
Using Degeneracy in the Loss Landscape for Mechanistic Interpretability
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Cindy Wu
Date:
December 14, 2025
Citations:
10
Secret Collusion Among Generative AI Agents
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Sumeet Motwani
Date:
December 14, 2025
Citations:
21
Distributed and Decentralised Training: Technical Governance Challenges in a Shifting AI Landscape
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Yashvardhan Sharma, Jakub Kryś
Date:
December 14, 2025
Citations:
0
Feature Hedging: Correlated Features Break Narrow Sparse Autoencoders
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
David Chanin
Date:
December 14, 2025
Citations:
4
Distillation Robustifies Unlearning
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Adeline (Addie) Foote, Alexander Infanger, Eleni Shor, Harish Kamath, Jacob Goldman-Wetzler
Date:
December 14, 2025
Citations:
4
The Attacker Moves Second: Stronger Adaptive Attacks Bypass Defenses Against Llm Jailbreaks and Prompt Injections
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Sander Schulhoff
Date:
December 14, 2025
Citations:
9
Narrow Finetuning Leaves Clearly Readable Traces in Activation Differences
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Clément Dumas, Julian Minder, Helena Casademunt
Date:
December 14, 2025
Citations:
0
BatchTopK Sparse Autoencoders
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Patrick Leask, Bart Bussmann
Date:
December 14, 2025
Citations:
48
Probabilistic Modeling of Latent Agentic Substructures in Deep Neural Networks
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Su Hyeong Lee
Date:
December 14, 2025
Citations:
0
Identifying Sparsely Active Circuits Through Local Loss Landscape Decomposition
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Brianna Chrisman
Date:
December 14, 2025
Citations:
2
Do the Rewards Justify the Means? Measuring Trade-Offs Between Rewards and Ethical Behavior in the MACHIAVELLI Benchmark
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Jonathan Ng, Hanlin Zhang
Date:
December 14, 2025
Citations:
163
Auditing language models for hidden objectives
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Florian Dietz, Kei Nishimura-Gasparian, Jeanne Salle
Date:
December 14, 2025
Citations:
22
Towards eliciting latent knowledge from LLMs with mechanistic interpretability
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Bartosz Cywiński, Emil Ryd
Date:
December 14, 2025
Citations:
5
On Defining Neural Averaging
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Su Hyeong Lee
Date:
December 14, 2025
Citations:
0
Transcoders Find Interpretable LLM Feature Circuits
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Jacob Dunefsky, Philippe Chlenski
Date:
December 14, 2025
Citations:
78
Too Late to Recall: The Two-Hop Problem in Multimodal Knowledge Retrieval
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Constantin Venhoff
Date:
December 14, 2025
Citations:
0
Convergent Linear Representations of Emergent Misalignment
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Ed Turner, Anna Soligo
Date:
December 14, 2025
Citations:
11
Understanding Reasoning in Thinking Language Models via Steering Vectors
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Constantin Venhoff
Date:
December 14, 2025
Citations:
38
Quantifying stability of non-power-seeking in artificial agents
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Evan Ryan Gunter, Yevgeny Liokumovich
Date:
December 14, 2025
Citations:
2
Understanding and Controlling a Maze-Solving Policy Network
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Ulisse Mini, Peli Grietzer
Date:
December 14, 2025
Citations:
20
Steering Language Models With Activation Engineering
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Lisa Thiergart, David Udell, Ulisse Mini
Date:
December 14, 2025
Citations:
243
Learning Multi-Level Features with Matryoshka Sparse Autoencoders
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Bart Bussmann, Adam Karvonen
Date:
December 14, 2025
Citations:
43
MISR: Measuring Instrumental Self-Reasoning in Frontier Models
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Kai Fronsdal
Date:
December 14, 2025
Citations:
1
Sparse Autoencoders Find Highly Interpretable Features in Language Models
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Fellow: Hoagy Cunningham
Date:
December 16, 2025
Citations:
738
Interpretability as Compression: Reconsidering SAE Explanations of Neural Activations with MDL-SAEs
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Kola Ayonrinde, Michael Pearce
Date:
December 14, 2025
Citations:
14
Evaluating Sparse Autoencoders on Targeted Concept Erasure Tasks
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Adam Karvonen, Can Rager
Date:
December 14, 2025
Citations:
8
Modular Training of Neural Networks aids Interpretability
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Satvik Golechha
Date:
December 14, 2025
Citations:
1
Audit Cards: Contextualizing AI Evaluations
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Leon Staufer, Mick Yang
Date:
December 14, 2025
Citations:
4
Teaching Models to Verbalize Reward Hacking in Chain-of-Thought Reasoning
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Marvin Li, Andy Arditi
Date:
December 14, 2025
Citations:
7
A Benchmark for Scalable Oversight Protocols
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Jackson Kaunismaa, Arjun Panickssery
Date:
December 14, 2025
Citations:
1
Language Models Learn to Mislead Humans via RLHF
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Jiaxin Wen
Date:
December 14, 2025
Citations:
69
Best-of-N Jailbreaking
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Sara Price, Aengus Lynch
Date:
December 14, 2025
Citations:
24
Representation Engineering: A Top-Down Approach to AI Transparency
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Shashwat Goel , Annah Dombrowski
Date:
December 14, 2025
Citations:
665
Future Events as Backdoor Triggers: Investigating Temporal Vulnerabilities in LLMs
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Arjun Panickssery, Sara Price
Date:
December 14, 2025
Citations:
12
Adaptive Deployment of Untrusted LLMs Reduces Distributed Threats
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Jiaxin Wen, Caleb Larson
Date:
December 14, 2025
Citations:
14
Among Us: A Sandbox for Measuring and Detecting Agentic Deception
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Satvik Golechha
Date:
December 14, 2025
Citations:
6
BadLlama: cheaply removing safety fine-tuning from Llama 2-Chat 13B
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Simon Lermen, Pranav Gade
Date:
December 14, 2025
Citations:
34
Refusal in Language Models Is Mediated by a Single Direction
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Aaquib Syed, Andy Arditi
Date:
December 14, 2025
Citations:
389
Training Dynamics of Contextual N-Grams in Language Models
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Lucia Quirke, Lovis Heindrich
Date:
December 14, 2025
Citations:
6
Limitations of Agents Simulated by Predictive Models
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Raymond Douglas, Jacek Karwowski, Chan Bae, Andis Draguns
Date:
December 14, 2025
Citations:
1
LLM Evaluators Recognize and Favor Their Own Generations
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Arjun Panickssery
Date:
December 14, 2025
Citations:
330
Finding Neurons in a Haystack: Case Studies with Sparse Probing
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Wes Gurnee
Date:
December 14, 2025
Citations:
275
Applying sparse autoencoders to unlearn knowledge in language models
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Eoin Farrell, Yeu-Tong Lau
Date:
December 14, 2025
Citations:
46
Goodhart's Law in Reinforcement Learning
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Jacek Karwowski
Date:
December 14, 2025
Citations:
12
Agentic Misalignment: How LLMs Could Be Insider Threats
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Caleb Larson, Aengus Lynch, Benjamin Wright
Date:
December 14, 2025
Citations:
31
Adaptive Sparse Allocation with Mutual Choice & Feature Choice Sparse Autoencoders
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Kola Ayonrinde
Date:
December 14, 2025
Citations:
7
On Targeted Manipulation and Deception when Optimizing LLMs for User Feedback
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Marcus Williams, Constantin Weisser
Date:
December 14, 2025
Citations:
40
Technical Report: Evaluating Goal Drift in Language Model Agents
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Elizabeth Donoway
Date:
December 14, 2025
Citations:
6
InterpBench: Semi-Synthetic Transformers for Evaluating Mechanistic Interpretability Techniques
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Rohan Gupta, Thomas Kwa
Date:
December 14, 2025
Citations:
5
Forecasting Frontier Language Model Agent Capabilities
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Govind Pimpale, Axel Højmark
Date:
December 14, 2025
Citations:
2
Measuring Progress in Dictionary Learning for Language Model Interpretability with Board Game Models
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Adam Karvonen, Can Rager, Benjamin Wright
Date:
December 14, 2025
Citations:
44
Constrained belief updates explain geometric structures in transformer representations
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Mateusz Piotrowski
Date:
December 14, 2025
Citations:
5
Steering Language Models with Weight Arithmetic
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Constanza Fierro
Date:
December 14, 2025
Citations:
0
Detecting Adversarial Fine-tuning with Auditing Agents
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Sarah Egler
Date:
December 14, 2025
Citations:
0
Is This the Subspace You Are Looking for? An Interpretability Illusion for Subspace Activation Patching
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Georg Lange, Aleksandar Makelov
Date:
December 14, 2025
Citations:
35
Will AI Tell Lies to Save Sick Children? Litmus-Testing AI Values Prioritization with AIRiskDilemmas
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Yu Ying Chiu, Sharan Maiya
Date:
December 14, 2025
Citations:
6
Me, Myself, and AI: The Situational Awareness Dataset (SAD) for LLMs
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Alexander Meinke, Rudolf Laine
Date:
December 14, 2025
Citations:
36
Model tampering attacks enable more rigorous evaluations of LLM capabilities
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Authors:
Zora Che
Date:
December 14, 2025
Citations:
21