An Approach to Technical AGI Safety and Security
This addresses safety and security concerns for AGI systems, which is a critical problem for humanity, but the approach is largely incremental, building on existing ideas in AI safety.
The paper tackles the risks of Artificial General Intelligence (AGI) by developing an approach to address misuse and misalignment risks, focusing on technical strategies like security measures, model-level mitigations, and system-level defenses to prevent harm to humanity.
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.