Imagen de Portada

Evaluating whether AI models would sabotage AI safety research

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Kirk, Robert, Souly, Alexandra, Fronsdal, Kai, D'Cruz, Abby, Davies, Xander
Formato:	Preprint
Publicado:	2026
Materias:	Artificial Intelligence
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

UK AISI Alignment Evaluation Case-Study
por: Souly, Alexandra, et al.
Publicado: (2026)

Breaking Agent Backbones: Evaluating the Security of Backbone LLMs in AI Agents
por: Bazinska, Julia, et al.
Publicado: (2025)

MISR: Measuring Instrumental Self-Reasoning in Frontier Models
por: Fronsdal, Kai, et al.
Publicado: (2024)

An Example Safety Case for Safeguards Against Misuse
por: Clymer, Joshua, et al.
Publicado: (2025)

Existing Large Language Model Unlearning Evaluations Are Inconclusive
por: Feng, Zhili, et al.
Publicado: (2025)

SeCodePLT: A Unified Platform for Evaluating the Security of Code GenAI
por: Nie, Yuzhou, et al.
Publicado: (2024)

Gram: Assessing sabotage propensities via automated alignment auditing
por: Lindner, David, et al.
Publicado: (2026)

Irresponsible AI: big tech's influence on AI research and associated impacts
por: Hernandez-Garcia, Alex, et al.
Publicado: (2025)

Empathy and the Right to Be an Exception: What LLMs Can and Cannot Do
por: Kidder, William, et al.
Publicado: (2024)

When Do LLM Preferences Predict Downstream Behavior?
por: Slama, Katarina, et al.
Publicado: (2026)

The implicated scientist: on the role of AI researchers in the development of weapons systems
por: Volokhova, Alexandra, et al.
Publicado: (2026)

How does information access affect LLM monitors' ability to detect sabotage?
por: Arike, Rauno, et al.
Publicado: (2026)

STACK: Adversarial Attacks on LLM Safeguard Pipelines
por: McKenzie, Ian R., et al.
Publicado: (2025)

Landscape of AI safety concerns -- A methodology to support safety assurance for AI-based autonomous systems
por: Schnitzer, Ronald, et al.
Publicado: (2024)

AgentHarm: A Benchmark for Measuring Harmfulness of LLM Agents
por: Andriushchenko, Maksym, et al.
Publicado: (2024)

AI Annotation Orchestration: Evaluating LLM verifiers to Improve the Quality of LLM Annotations in Learning Analytics
por: Ahtisham, Bakhtawar, et al.
Publicado: (2025)

Security Challenges in AI Agent Deployment: Insights from a Large Scale Public Competition
por: Zou, Andy, et al.
Publicado: (2025)

What Is AI Safety? What Do We Want It to Be?
por: Harding, Jacqueline, et al.
Publicado: (2025)

Deep Ignorance: Filtering Pretraining Data Builds Tamper-Resistant Safeguards into Open-Weight LLMs
por: O'Brien, Kyle, et al.
Publicado: (2025)

Fundamental Limitations in Pointwise Defences of LLM Finetuning APIs
por: Davies, Xander, et al.
Publicado: (2025)

A Case for AI Consciousness: Language Agents and Global Workspace Theory
por: Goldstein, Simon, et al.
Publicado: (2024)

Is the future of AI green? What can innovation diffusion models say about generative AI's environmental impact?
por: Viseur, Robert, et al.
Publicado: (2026)

"Just a strange pic": Evaluating 'safety' in GenAI Image safety annotation tasks from diverse annotators' perspectives
por: Wang, Ding, et al.
Publicado: (2025)

Towards evaluations-based safety cases for AI scheming
por: Balesni, Mikita, et al.
Publicado: (2024)

Developing and Maintaining an Open-Source Repository of AI Evaluations: Challenges and Insights
por: Abbas, Alexandra, et al.
Publicado: (2025)

SAGE: Scalable AI Governance & Evaluation
por: Le, Benjamin, et al.
Publicado: (2026)

AI-Assisted Systematization for Evaluating GenAI Systems
por: Agarwal, Dhruv, et al.
Publicado: (2026)

A sketch of an AI control safety case
por: Korbak, Tomek, et al.
Publicado: (2025)

Affirmative safety: An approach to risk management for high-risk AI
por: Wasil, Akash R., et al.
Publicado: (2024)

Towards provable probabilistic safety for scalable embodied AI systems
por: He, Linxuan, et al.
Publicado: (2025)

Streamlining the review process: AI-generated annotations in research manuscripts
por: Díaz, Oscar, et al.
Publicado: (2024)

Sustaining AI safety: Control-theoretic external impossibility, intrinsic necessity, and structural requirements
por: Mazzu, James M.
Publicado: (2026)

Position: AI Competitions Provide the Gold Standard for Empirical Rigor in GenAI Evaluation
por: Sculley, D., et al.
Publicado: (2025)

Lean-ing on Quality: How High-Quality Data Beats Diverse Multilingual Data in AutoFormalization
por: Chan, Willy, et al.
Publicado: (2025)

Trustworthy, responsible, ethical AI in manufacturing and supply chains: synthesis and emerging research questions
por: Brintrup, Alexandra, et al.
Publicado: (2023)

Toward best research practices in AI Psychology
por: Ivanova, Anna A.
Publicado: (2023)

TPS-Bench: Evaluating AI Agents' Tool Planning \& Scheduling Abilities in Compounding Tasks
por: Xu, Hanwen, et al.
Publicado: (2025)

Seven simple steps for log analysis in AI systems
por: Dubois, Magda, et al.
Publicado: (2026)

A cross-regional review of AI safety regulations in the commercial aviation
por: Barr, Penny A., et al.
Publicado: (2025)

The AI Literacy Heptagon: A Structured Approach to AI Literacy in Higher Education
por: Hackl, Veronika, et al.
Publicado: (2025)