A/HRC/43/29 D. Ensuring accountability 50. Accountability for human rights violations is central to the framework of international human rights obligations. This framework defines who is responsible for what and towards whom, and expresses the obligations assumed by States to take steps, individually and through international assistance and cooperation, especially economic and technical, to the maximum of available resources, with a view to achieving progressively the full realization of economic, social and cultural rights. It makes clear that certain obligations are of an immediate nature, notably to remove discriminatory laws, policies and other measures and to assure minimum essential levels of each right for all, with particular attention to those left furthest behind. The framework enables duty bearers to be held accountable to rights holders for their decisions or omissions, and provides mechanisms for claiming rights, monitoring progress transparently, sanctioning poor performance and seeking redress for human rights violations. 51. While new, data-driven technologies bring about new challenges in terms of ensuring human rights accountability, a variety of tools and safeguarding methodologies exist to identify and address risks and harms. Appropriate due diligence processes, taking into account the full range of rights under international human rights law throughout the life cycle of a technological system, can help avoid unduly narrow analysis of potential risks. Such processes can be helpful in identifying and preventing possible human rights harm, including by determining necessary safeguards, and in developing effective remedies when harm does occur. Meaningful consultations with external stakeholders and, where possible, with representatives of potentially impacted individuals and groups, in order to avoid project-driven bias, can strengthen such processes and significantly enhance their effectiveness (A/73/348, para. 54). On this basis, it would, for example, be recommended to integrate ongoing human rights diligence and broad consultations into the process of developing and deploying comprehensive nationwide digital identification systems, in order to enable the identification and mitigation of human rights risks associated with the systems. 52. Often, there may be significant gaps in public knowledge and understanding concerning the technological means being used by Governments and private actors in many public services, such as social security, pensions, health care, taxation, education or recruitment. This is a particular problem in the context of the automated decision-making processes that rely on artificial intelligence. Comprehensive, publicly available information is important to enable informed decision-making and the relevant consent of affected parties. It is advisable to require administrative services to systematically inform the addressees of rights-affecting decisions if those decisions have been made automatically or with the help of automation tools. For human rights-critical applications, the introduction of registers containing key information about those tools and their use can be considered. Regulations requiring companies to disclose when artificial intelligence systems are used in ways that affect the exercise of human rights and share the results of related human rights impact assessments may also be a helpful tool. 53. An associated dimension in the use of artificial intelligence technologies is “explainability”, concerned with the tendency towards opacity of complex algorithmic tools, also known as the “black box” problem. Such systems, in particular those with selflearning capabilities, can often behave in a fashion that is not entirely explainable or predictable. On occasion, intellectual property protections may, in this context, prevent necessary scrutiny of the algorithms and data used to train them. Even access to the underlying source code and training data, however, may often be insufficient to offer adequate understanding of how a particular artificial intelligence system operates in practice. Additional efforts are necessary to create tools and methods that provide a sufficient level of explanation of how decisions have been reached, in particular when artificial intelligence is determining critical issues within judicial processes or regarding access to, eligibility for and use of critical social services that are essential for the realization of economic, social and cultural rights. 54. In addition, regular audits by internal and external reviewers throughout the life cycle of artificial intelligence systems can provide a critical guarantee of rigour and independence 12