A/HRC/44/57 to combat racist and xenophobic speech and conduct, including online.116 These obligations apply equally to the issues analysed in the present report: in the context of emerging digital technologies, States must take effective measures to detect and combat racially discriminatory design and use of such technologies in access to civil, political, economic, social and cultural rights.117 55. States’ obligations to prevent and eliminate racial discrimination in the design and use of emerging digital technologies require addressing the “diversity crisis” in the sectors discussed in part II above. States must work together with private corporations, including on the basis of legally binding frameworks, to develop the necessary special measures to ensure that racial and ethnic minorities are meaningfully represented in all aspects of decision-making relating to the design and use of emerging digital technologies. A genuine shift in power is required in the various sectors of emerging digital technologies, and not merely tokenism of women and racial and ethnic minority groups. Central to shifting power – even within the private sector – are deeper engagement with and funding for research and knowledge production that specifically aim to deepen understanding of discrimination in the design and use of emerging digital technologies from an interdisciplinary perspective. Researchers affiliated with the Center for Critical Race and Digital Studies offer examples.118 56. States must take swift and effective action to prevent and mitigate the risk of the racially discriminatory use and design of emerging digital technologies, including by making racial equality and non-discrimination human rights impact assessments a prerequisite for the adoption of systems based on such technologies by public authorities. These impact assessments must incorporate meaningful opportunity for co-design and co-implementation with representatives of racially or ethnically marginalized groups. A purely or even mainly voluntary approach to equality impact assessments will not suffice; a mandatory approach is essential. Recent developments in this direction from the Council of Europe,119 for example, are laudable. They must neither neglect racial discrimination nor exclude racial and ethnic minorities from decision-making. Sometimes prevention of racially discriminatory outcomes and other human rights violations in the design and use of emerging digital technologies by public authorities may require outright bans on their use until the risk of their harm is sufficiently mitigated. The city of San Francisco’s decision to ban local government use of facial recognition software is a good example in this regard. 57. In order to comply with their equality and non-discrimination obligations, States must ensure transparency and accountability for public sector use of emerging digital technologies, and enable independent analysis and oversight, including by only using systems that are auditable. Recent reforms by Canada to implement public sector accountability for the use of emerging digital technologies provide examples of an important first step in this regard.120 58. States must ensure that ethical frameworks and guidelines developed to provide flexible, practical and effective regulation and governance of emerging digital technologies are grounded in legally binding international human rights principles, including those prohibiting racial discrimination. The Toronto Declaration on protecting the right to equality and non-discrimination in machine learning systems exemplifies the symbiotic relationship that should exist between binding international human rights law and ethical guidelines or principles for artificial intelligence governance.121 The Toronto Declaration stresses the binding nature of equality and non116 117 118 119 120 121 See A/73/305, A/73/312 and A/HRC/38/53. Article 5 of the Convention makes explicit the application of the prohibition on racial discrimination to access and enjoyment of civil, political, social, economic and cultural rights. See https://criticalracedigitalstudies.com. Recommendation CM/Rec(2020)1 of the Committee of Ministers to member States on the human rights impacts of algorithmic systems, 8 April 2020. See www.tbs-sct.gc.ca/pol/doc-eng.aspx?id=32592. See www.torontodeclaration.org/declaration-text/english. 17