US Intelligence Community releases AI principles

The US Intelligence Community (IC) has released a set of Principles of Artificial Intelligence Ethics and an AI Ethics Framework to guide the IC’s ethical development and use of artificial intelligence (AI). The Principles for AI Ethics, intended to ‘guide personnel of whether and how to develop and use AI’, include: (a) respecting the law and acting with integrity; (b) being transparent and accountable; (c) being objective and equitable; (d) being human-centred in development and use; (e) being secure and resilient; (e) being informed by science and technology. The AI Ethics Framework includes guidelines for the US IC personnel on how to procure, design, build, use, protect, consume, and manage AI and related data. Elements outlined in the framework include: (a) understanding goals and risks; (b) understanding legal obligations and policy considerations governing AI and data; (c) applying human judgement and accountability; (d) mitigating undesired bias and ensuring objectivity; (e) testing AI systems for accuracy; (f) accounting for builds, versions, and evolution of an AI system; (g) documenting purpose, parameters, limitations, and design outcomes; (h) using methods that are explainable and understandable; (i) undertaking reviews to determine whether the AI system still meets its purpose and whether undesired biases or unintended outcomes are appropriately mitigated; (j) ensuring clarity on who has the responsibility for the continued maintenance, monitoring, updating, and decommissioning of the AI system.

Related issues