English
Related papers

Related papers: Training Deliberative Monitors for Black-Box Schem…

200 papers

Highly capable AI systems could secretly pursue misaligned goals -- what we call "scheming". Because a scheming AI would deliberately try to hide its misaligned goals and actions, measuring and mitigating scheming requires different…

Safe deployment of Large Language Model (LLM) agents in autonomous settings requires reliable oversight mechanisms. A central challenge is detecting scheming, where agents covertly pursue misaligned goals. One approach to mitigating such…

Computation and Language · Computer Science 2026-03-03 Simon Storf , Rich Barton-Cooper , James Peters-Gill , Marius Hobbhahn

Observability into the decision making of modern AI systems may be required to safely deploy increasingly capable agents. Monitoring the chain-of-thought (CoT) of today's reasoning models has proven effective for detecting misbehavior.…

Frontier AI agents may pursue hidden goals while concealing their pursuit from oversight. Alignment training aims to prevent such behavior by reinforcing the correct goals, but alignment may not always succeed and can lead to unwanted side…

Machine Learning · Computer Science 2026-02-27 Bruce W. Lee , Chen Yueh-Han , Tomek Korbak

Frontier models are increasingly trained and deployed as autonomous agent. One safety concern is that AI agents might covertly pursue misaligned goals, hiding their true capabilities and objectives - also known as scheming. We study whether…

Artificial Intelligence · Computer Science 2025-01-16 Alexander Meinke , Bronson Schoen , Jérémy Scheurer , Mikita Balesni , Rusheb Shah , Marius Hobbhahn

Recent work has demonstrated the plausibility of frontier AI models scheming -- knowingly and covertly pursuing an objective misaligned with its developer's intentions. Such behavior could be very hard to detect, and if present in future…

Machine Learning · Computer Science 2025-07-04 Mary Phuong , Roland S. Zimmermann , Ziyue Wang , David Lindner , Victoria Krakovna , Sarah Cogan , Allan Dafoe , Lewis Ho , Rohin Shah

As large language model (LLM) agents are deployed autonomously in diverse contexts, evaluating their capacity for strategic deception becomes crucial. While recent research has examined how AI systems scheme against human developers,…

Computation and Language · Computer Science 2026-04-28 Thao Pham

Mitigating reward hacking--where AI systems misbehave due to flaws or misspecifications in their learning objectives--remains a key challenge in constructing capable and aligned models. We show that we can monitor a frontier reasoning…

Artificial Intelligence · Computer Science 2025-03-18 Bowen Baker , Joost Huizinga , Leo Gao , Zehao Dou , Melody Y. Guan , Aleksander Madry , Wojciech Zaremba , Jakub Pachocki , David Farhi

Monitoring the chain-of-thought (CoT) of reasoning models is a promising approach for detecting covert misbehavior (i.e., hidden objectives) in code generation tasks. While large models (GPT-5, Gemini-3-Flash) can serve as effective CoT…

Cryptography and Security · Computer Science 2026-05-14 Nirav Diwan , Han Wang , Berkcan Kapusuzoglu , Ramin Moradi , Supriyo Chakraborty , Giri Iyengar , Sambit Sahu , Huan Zhang , Gang Wang

Strategic model selection and reasoning settings are more effective than ensembling for optimizing automated scoring with large language models (LLMs). We examined self-consistency (intra-model majority voting) and reasoning effort for…

Computers and Society · Computer Science 2026-05-01 Scott Frohn

Frontier language models are deployed as black-box services, where model weights cannot be modified and customization is limited to prompting. We introduce Advisor Models, a method to train small open-weight models to generate dynamic,…

Machine Learning · Computer Science 2026-05-18 Parth Asawa , Alan Zhu , Abigail O'Neill , Matei Zaharia , Alexandros G. Dimakis , Joseph E. Gonzalez

As AI systems are increasingly deployed in autonomous agentic settings at scale, it is important to ensure the actions they take are safe and aligned with user intent. Monitoring agent actions is a key safety mechanism, yet reliable…

Artificial Intelligence · Computer Science 2026-05-19 Eugene Koran , Yejun Yun , Samantha Tetef , Benjamin Arnav , Pablo Bernabeu-Pérez

Current LLM safety defenses fail under decomposition attacks, where a malicious goal is decomposed into benign subtasks that circumvent refusals. The challenge lies in the existing shallow safety alignment techniques: they only detect harm…

Cryptography and Security · Computer Science 2025-06-17 Chen Yueh-Han , Nitish Joshi , Yulin Chen , Maksym Andriushchenko , Rico Angell , He He

Open reasoning language models are often compared under mixed sample sizes, partially standardized prompts, and accuracy-centered summaries, which makes practical model selection difficult to interpret. We present a unified evaluation of…

Computation and Language · Computer Science 2026-05-20 Md Motaleb Hossen Manik , Ge Wang

Large reasoning models (LRMs) such as Claude 3.7 Sonnet and OpenAI o1 achieve strong performance on mathematical benchmarks using lengthy chain-of-thought (CoT) reasoning, but the resulting traces are often unnecessarily verbose. This…

Computation and Language · Computer Science 2025-06-13 Ye Yu , Yaoning Yu , Haohan Wang

As agentic coding systems decompose work across multiple model instances, a critical safety question is whether those instances can coordinate to achieve a hidden malicious objective while remaining aligned with user intent. We introduce…

Cryptography and Security · Computer Science 2026-05-29 Nikolay Radev , Lennart Haas , Benjamin Arnav , Pablo Bernabeu-Pérez

As Large Language Models (LLMs) are increasingly deployed as autonomous agents in complex and long horizon settings, it is critical to evaluate their ability to sabotage users by pursuing hidden objectives. We study the ability of frontier…

As agentic AI systems increasingly operate autonomously, establishing trust through verifiable evaluation becomes critical. Yet existing benchmarks lack the transparency and auditability needed to assess whether agents behave reliably. We…

Computation and Language · Computer Science 2025-12-02 Hyunjun Kim , Sooyoung Ryu

Machine learning (ML) systems have achieved remarkable performance across a wide area of applications. However, they frequently exhibit unfair behaviors in sensitive application domains, raising severe fairness concerns. To evaluate and…

Software Engineering · Computer Science 2024-07-02 Yisong Xiao , Aishan Liu , Tianlin Li , Xianglong Liu

This study presents a comprehensive, long-term project to explore the effectiveness of various prompting techniques in detecting dialogical mental manipulation. We implement Chain-of-Thought prompting with Zero-Shot and Few-Shot settings on…

Computation and Language · Computer Science 2024-08-15 Ivory Yang , Xiaobo Guo , Sean Xie , Soroush Vosoughi
‹ Prev 1 2 3 10 Next ›