An Approach to Technical AGI Safety and Security

• URL: http://arxiv.org/abs/2504.01849v1
• Date: Wed, 02 Apr 2025 15:59:31 GMT
• Title: An Approach to Technical AGI Safety and Security
• Authors: Rohin Shah, Alex Irpan, Alexander Matt Turner, Anna Wang, Arthur Conmy, David Lindner, Jonah Brown-Cohen, Lewis Ho, Neel Nanda, Raluca Ada Popa, Rishub Jain, Rory Greig, Samuel Albanie, Scott Emmons, Sebastian Farquhar, Sébastien Krier, Senthooran Rajamanoharan, Sophie Bridgers, Tobi Ijitoye, Tom Everitt, Victoria Krakovna, Vikrant Varma, Vladimir Mikulik, Zachary Kenton, Dave Orr, Shane Legg, Noah Goodman, Allan Dafoe, Four Flynn, Anca Dragan,
• Abstract summary: We develop an approach to address the risk of harms consequential enough to significantly harm humanity.<n>We focus on technical approaches to misuse and misalignment.<n>We briefly outline how these ingredients could be combined to produce safety cases for AGI systems.
• Score: 72.83728459135101
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Artificial General Intelligence (AGI) promises transformative benefits but also presents significant risks. We develop an approach to address the risk of harms consequential enough to significantly harm humanity. We identify four areas of risk: misuse, misalignment, mistakes, and structural risks. Of these, we focus on technical approaches to misuse and misalignment. For misuse, our strategy aims to prevent threat actors from accessing dangerous capabilities, by proactively identifying dangerous capabilities, and implementing robust security, access restrictions, monitoring, and model safety mitigations. To address misalignment, we outline two lines of defense. First, model-level mitigations such as amplified oversight and robust training can help to build an aligned model. Second, system-level security measures such as monitoring and access control can mitigate harm even if the model is misaligned. Techniques from interpretability, uncertainty estimation, and safer design patterns can enhance the effectiveness of these mitigations. Finally, we briefly outline how these ingredients could be combined to produce safety cases for AGI systems.

Related papers

• Concept Enhancement Engineering: A Lightweight and Efficient Robust Defense Against Jailbreak Attacks in Embodied AI [19.094809384824064]
  Embodied Intelligence (EI) systems integrated with large language models (LLMs) face significant security risks. Traditional defense strategies, such as input filtering and output monitoring, often introduce high computational overhead. We propose Concept Enhancement Engineering (CEE) to enhance the safety of embodied LLMs by dynamically steering their internal activations.
  arXiv  Detail & Related papers  (2025-04-15T03:50:04Z)
• Representation Bending for Large Language Model Safety [27.842146980762934]
  Large Language Models (LLMs) have emerged as powerful tools, but their inherent safety risks pose significant challenges.<n>This paper introduces RepBend, a novel approach that fundamentally disrupts the representations underlying harmful behaviors in LLMs.<n>RepBend achieves state-of-the-art performance, outperforming prior methods such as Circuit Breaker, RMU, and NPO, with up to 95% reduction in attack success rates.
  arXiv  Detail & Related papers  (2025-04-02T09:47:01Z)
• Safety and Security Risk Mitigation in Satellite Missions via Attack-Fault-Defense Trees [2.252059459291148]
  This work presents a case study from Ascentio Technologies, a mission-critical system company in Argentina specializing in aerospace. The main focus will be on the Ground Segment for the satellite project currently developed by the company. This paper showcases the application of the Attack-Fault-Defense Tree framework, which integrates attack trees, fault trees, and defense mechanisms into a unified model.
  arXiv  Detail & Related papers  (2025-04-01T17:24:43Z)
• Open Problems in Machine Unlearning for AI Safety [61.43515658834902]
  Machine unlearning -- the ability to selectively forget or suppress specific types of knowledge -- has shown promise for privacy and data removal tasks. In this paper, we identify key limitations that prevent unlearning from serving as a comprehensive solution for AI safety.
  arXiv  Detail & Related papers  (2025-01-09T03:59:10Z)
• Defining and Evaluating Physical Safety for Large Language Models [62.4971588282174]
  Large Language Models (LLMs) are increasingly used to control robotic systems such as drones. Their risks of causing physical threats and harm in real-world applications remain unexplored. We classify the physical safety risks of drones into four categories: (1) human-targeted threats, (2) object-targeted threats, (3) infrastructure attacks, and (4) regulatory violations.
  arXiv  Detail & Related papers  (2024-11-04T17:41:25Z)
• EARBench: Towards Evaluating Physical Risk Awareness for Task Planning of Foundation Model-based Embodied AI Agents [53.717918131568936]
  Embodied artificial intelligence (EAI) integrates advanced AI models into physical entities for real-world interaction.<n>Foundation models as the "brain" of EAI agents for high-level task planning have shown promising results.<n>However, the deployment of these agents in physical environments presents significant safety challenges.<n>This study introduces EARBench, a novel framework for automated physical risk assessment in EAI scenarios.
  arXiv  Detail & Related papers  (2024-08-08T13:19:37Z)
• Towards Guaranteed Safe AI: A Framework for Ensuring Robust and Reliable AI Systems [88.80306881112313]
  We will introduce and define a family of approaches to AI safety, which we will refer to as guaranteed safe (GS) AI. The core feature of these approaches is that they aim to produce AI systems which are equipped with high-assurance quantitative safety guarantees. We outline a number of approaches for creating each of these three core components, describe the main technical challenges, and suggest a number of potential solutions to them.
  arXiv  Detail & Related papers  (2024-05-10T17:38:32Z)
• Safety Margins for Reinforcement Learning [53.10194953873209]
  We show how to leverage proxy criticality metrics to generate safety margins. We evaluate our approach on learned policies from APE-X and A3C within an Atari environment.
  arXiv  Detail & Related papers  (2023-07-25T16:49:54Z)
• Visual Adversarial Examples Jailbreak Aligned Large Language Models [66.53468356460365]
  We show that the continuous and high-dimensional nature of the visual input makes it a weak link against adversarial attacks. We exploit visual adversarial examples to circumvent the safety guardrail of aligned LLMs with integrated vision. Our study underscores the escalating adversarial risks associated with the pursuit of multimodality.
  arXiv  Detail & Related papers  (2023-06-22T22:13:03Z)
• Sustainable Adaptive Security [11.574868434725117]
  We propose the notion of Sustainable Adaptive Security (SAS) which reflects enduring protection by augmenting adaptive security systems with the capability of mitigating newly discovered threats. We use a smart home example to showcase how we can engineer the activities of the MAPE (Monitor, Analysis, Planning, and Execution) loop of systems satisfying sustainable adaptive security.
  arXiv  Detail & Related papers  (2023-06-05T08:48:36Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.