‘Handle with Care’: Due Diligence Obligations in the Employment of AI Technologies

While artificial intelligence (AI) has enormous potential for good, it also gives rise to significant risks. International law applies by default to this technology. It imposes on states not only negative obligations to refrain from engaging in certain AI-related risks or harms but also positive duties to prevent those from arising in the first place. This Chapter unpacks those positive obligations requiring states to exercise due diligence in the design, development, and deployment of AI. It argues that due diligence is a standard of conduct found in different ‘protective’ obligations. The Chapter then zooms into some of those rules, including so-called ‘Corfu Channel’ and ‘no-harm’ principles, and positive duties under international humanitarian and human rights law. It concludes by recommending specific due diligence measures applicable to AI, such as impact assessments, testing and training, continued oversight, human involvement, and regulatory measures.