A/HRC/56/68 despite the clarity and timeliness of prior calls for a “colour-blind” approach and the increase in awareness of systemic racism in the intervening four years. The assumption that technology is objective and neutral remains pervasive and drives a race to integrate artificial intelligence into society despite its racially discriminatory impacts and without due consideration of whether it is necessary. While artificial intelligence does have positive potential, including for equality and inclusion, it is not a panacea for all societal issues and must be effectively managed to balance its benefits and risks. 67. The effective and comprehensive regulation of artificial intelligence is central to achieving this careful balance. While the effective regulation of artificial intelligence is vital, there are additional steps that States and others can take to effectively address the racially discriminatory impacts of these technologies. Developing human rights-based public education about emerging technologies and building artificial intelligence literacy are also very important. When individuals and groups understand artificial intelligence and are aware of their human rights in the digital space, they are empowered to use that knowledge responsibly and become a discerning and responsible audience that can improve accountability for artificial intelligence systems. 68. States should: (a) Address the challenge of regulating artificial intelligence with a greater sense of urgency, bearing in mind the speed with which these technologies are being developed and the multitude of ways in which they are already perpetuating racial discrimination across societal domains; (b) Develop artificial intelligence regulatory frameworks that are based on a comprehensive understanding of systemic racism and are grounded in international human rights law, including the prohibition of racial discrimination. Such frameworks should not be based on siloed approaches and should take into account different legal instruments, including dedicated artificial intelligence legislation, privacy laws, freedom of information provisions, anti-discrimination legislation and sectoral regulations, to achieve comprehensive and effective regulation that prevents and addresses the racially discriminatory impact of artificial intelligence; (c) Consider the role that voluntary standards can play in artificial intelligence regulatory frameworks. Voluntary standards can provide actionable guidelines on the practical implementation of legal standards. However, artificial intelligence regulation should not rely solely on voluntary standards, due to the significance of the human rights implications of these technologies, including in relation to racial discrimination; (d) Enshrine a legally binding obligation, within regulatory frameworks, to conduct comprehensive human rights due diligence assessments, including explicit criteria to assess racial and ethnic bias, in the development and deployment of all artificial intelligence technologies. Human rights due diligence should include data-testing protocols and thresholds that safeguard against algorithmic bias, including racial and ethnic bias. They should be completed before the deployment of new technologies, particularly in public settings, such as educational, law enforcement and health-care settings; (e) Consider prohibiting the use of artificial intelligence systems that have been shown to have unacceptable human rights risks, including those that violate the prohibition of racial discrimination; (f) Ensure that there are provisions within regulatory frameworks to guarantee full transparency about automated decision-making processes, including rights of access to information, in cases in which artificial intelligence use is deemed permissible, based on comprehensive human rights due diligence; (g) Put in place clear and accessible appeals processes, which have a mandate to assess and address the racially discriminatory impacts of artificial intelligence and involve human review. Equitable access to such appeals processes should be ensured; 18 GE.24-08849