mlsanityData to 7 Oct 2026

Evaluating Frontier Models for Dangerous Capabilities

arXiv 2403.13793 ↗First posted 20 Mar 2024cs.LGcited by 121

Not on the map. Why?

Cited by 29 in LLM agents and tool use, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

To understand the risks posed by a new AI system, we must understand what it can and cannot do. Building on prior work, we introduce a programme of new "dangerous capability" evaluations and pilot them on Gemini 1.0 models. Our evaluations cover four areas: (1) persuasion and deception; (2) cyber-security; (3) self-proliferation; and (4) self-reasoning. We do not find evidence of strong dangerous capabilities in the models we evaluated, but we flag early warning signs. Our goal is to help advance a rigorous science of dangerous capability evaluation, in preparation for future models.

Key facts. Cited by 121 papers in the map (data through 7 Oct 2026). First posted on arXiv 20 Mar 2024.

Builds on

0 of 0 references.

No references here yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 45; 2025: 47; 2026: 29.

121 papers.

Sort:

By area:

  1. 2024OpenAI o1 System Cardcited by 1,220
  2. 2025Humanity's Last Examcited by 467
  3. 2024Frontier Models are Capable of In-context Schemingcited by 252
  4. 2025Measuring AI Ability to Complete Long Software Taskscited by 142
  5. 2024AI Sandbagging: Language Models can Strategically Underperform on Evaluationscited by 131
  6. 2024LLM Agents can Autonomously Exploit One-day Vulnerabilitiescited by 125
  7. 2025Measuring what Matters: Construct Validity in Large Language Model Benchmarkscited by 102
  8. 2024Best-of-N Jailbreakingcited by 71
  9. 2024Refusal-Trained LLMs Are Easily Jailbroken As Browser Agentscited by 61
  10. 2024Introducing v0.5 of the AI Safety Benchmark from MLCommonscited by 60
  11. 2025SHADE-Arena: Evaluating Sabotage and Monitoring in LLM Agentscited by 58
  12. 2024EnIGMA: Interactive Tools Substantially Assist LM Agents in Finding Security Vulnerabilitiescited by 51
38 more
  1. 2024LatentQA: Teaching LLMs to Decode Activations Into Natural Languagecited by 49
  2. 2024Surgical, Cheap, and Flexible: Mitigating False Refusal in Language Models via Single Vector Ablationcited by 48
  3. 2024Sabotage Evaluations for Frontier Modelscited by 45
  4. 2024Stress-Testing Capability Elicitation With Password-Locked Modelscited by 42
  5. 2025A sketch of an AI control safety casecited by 36
  6. 2024Towards evaluations-based safety cases for AI schemingcited by 32
  7. 2024Adversaries Can Misuse Combinations of Safe Modelscited by 31
  8. 2025Evaluating Frontier Models for Stealth and Situational Awarenesscited by 30
  9. 2025Representation Bending for Large Language Model Safetycited by 28
  10. 2024Towards interactive evaluations for interaction harms in human-AI systemscited by 27
  11. 2026The 2025 AI Agent Index: Documenting Technical and Safety Features of Deployed Agentic AI Systemscited by 27
  12. 2025Reliable Weak-to-Strong Monitoring of LLM Agentscited by 24
  13. 2026ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks?cited by 24
  14. 2025Nuclear Deployed: Analyzing Catastrophic Risks in Decision-making of Autonomous LLM Agentscited by 23
  15. 2025Lessons From Red Teaming 100 Generative AI Productscited by 22
  16. 2025Cyber-Zero: Training Cybersecurity Agents without Runtimecited by 22
  17. 2025Frontier AI's Impact on the Cybersecurity Landscapecited by 21
  18. 2025How to evaluate control measures for LLM agents? A trajectory from today to superintelligencecited by 20
  19. 2025PropensityBench: Evaluating Latent Safety Risks in Large Language Models via an Agentic Approachcited by 19
  20. 2024Holistic Safety and Responsibility Evaluations of Advanced AI Modelscited by 18
  21. 2024Breach By A Thousand Leaks: Unsafe Information Leakage in `Safe' AI Responsescited by 18
  22. 2025LLMs Can Covertly Sandbag on Capability Evaluations Against Chain-of-Thought Monitoringcited by 17
  23. 2024Can LLMs make trade-offs involving stipulated pain and pleasure states?cited by 16
  24. 2024Frontier AI systems have surpassed the self-replicating red linecited by 16
  25. 2025A Framework for Evaluating Emerging Cyberattack Capabilities of AIcited by 15
  26. 2025Training Language Model Agents to Find Vulnerabilities with CTF-Dojocited by 15
  27. 2024From Imitation to Introspection: Probing Self-Consciousness in Language Modelscited by 14
  28. 2024Hacking CTFs with Plain Agentscited by 14
  29. 2024Lies, Damned Lies, and Distributional Language Statistics: Persuasion and Deception with Large Language Modelscited by 14
  30. 2025The AI Agent Indexcited by 14
  31. 2024Training Compute Thresholds: Features and Functions in AI Regulationcited by 13
  32. 2026Measuring AI Agents' Progress on Multi-Step Cyber Attack Scenarioscited by 13
  33. 2024Catastrophic Cyber Capabilities Benchmark (3CB): Robustly Evaluating LLM Agent Cyber Offense Capabilitiescited by 12
  34. 2025OCCULT: Evaluating Large Language Models for Offensive Cyber Operation Capabilitiescited by 12
  35. 2025Must Read: A Comprehensive Survey of Computational Persuasioncited by 12
  36. 2025Evaluation Faking: Unveiling Observer Effects in Safety Evaluation of Frontier AI Systemscited by 12
  37. 2025Scaling Up Active Testing to Large Language Modelscited by 12
  38. 2025Can AI Perceive Physical Danger and Intervene?cited by 12

50 most cited of 121 All 121 (JSON)