Guardrails and coordination on AI: assessing the HRC’s NET Resolution Renewal

During the recent 53rd session of the UN Human Rights Council (HRC), states convened to discuss the renewal of resolution 41/11—a landmark instrument passed in 2019, which called on the HRC’s Advisory Committee to produce and present a report identifying the impacts, opportunities and challenges presented by new and emerging technologies (NETs) and their intersections with human rights. 

Published in 2021, the report focused on mapping and assessing the impacts of NETs, and outlined best practices at the national and regional levels. The report also called for a systematic review of human rights and NETs in the UN context, a request that was ultimately included in the subsequent 2021 iteration of the resolution, and led to the UN Office of the High Commissioner on Human Rights’ (OHCHR) just published report on technical standards and human rights

At this most recent HRC session, discussions around the resolution moved from diagnosis to a more substantive approach. Participating states, led by the ‘Core Group’ who co-sponsored the resolution (Austria, Brazil, Denmark, Morocco, Singapore, and South Korea), proposed a suite of new, amended and expanded provisions within the resolution, with wide-ranging implications for its ongoing implementation. GPD provided informal written input into the process, and closely monitored the negotiations alongside civil society partners. 

Below, we summarise and unpack the key changes in the 2023 version to the resolution, as well as their likely implications for human rights in the context of emerging technologies.

New focus on AI

In discussions on the resolution’s renewal, the Core Group chose to focus on artificial intelligence (AI): harnessing its possible benefits for human rights while avoiding negative impacts through appropriate safeguards.

The renewed resolution now emphasises the importance of ensuring human rights safeguards are integrated throughout the lifecycle of artificial intelligence systems—a key win for human rights, and something for which GPD and other civil society organisations have consistently advocated.  The renewed resolution also includes important new recommendations for ensuring human oversight and accountability in relation to uses of artificial intelligence systems. However, it’s regrettable that more explicit language on ensuring that these safeguards are introduced prior to the dissemination and implementation of artificial intelligence systems was not ultimately included.

It is also important to recognise the areas where the resolution could and should have gone further—such as its failure to include a recommendation to halt or ban the usage and/or dissemination of artificial intelligence systems whose applications are incompatible with existing international human rights law. This is something for which GPD and other civil society groups such as Article 19 advocated, and would have reflected the recommendations of the OHCHR’s 2021 report ‘The right to privacy in the digital age’. The final compromise was a weakened version, acknowledging that “certain applications of artificial intelligence present an unacceptable risk to human rights”, and citing examples such as “behavioural prediction or the scoring of individuals” without following the recommendation to establish a red line. 

Stronger focus on vulnerable and underrepresented groups

The renewed resolution expands on the category of ‘vulnerable’ groups, while acknowledging that they require additional safeguards. New categories of groups are referenced in the resolution, such as indigenous peoples, persons in rural areas, economically disadvantaged persons, and persons in vulnerable or marginalised situations. The drafters also added language on considering gender in AI systems, including noting the links between AI systems and sexual and gender-based violence and how NETs can amplify the impact of such attacks. However, the resolution could have gone further by explicitly referring to people within a migration context—including undocumented migrants, refugees, asylum-seekers, internally displaced and stateless people—who often face disproportionate impacts from AI systems used in border control settings.

In a win for human rights, the Core Group added references to the importance of ensuring that the usage of AI systems is aligned with the protection, promotion and enjoyment of both civil and political rights—including the right to privacy and freedom of opinion and expression—and economic, social, and cultural (ECOSOC) rights. The significance of the resolution’s recognition of the importance of ECOSOC rights cannot be understated, given the impacts of AI dissemination and usage on discrimination and inequality. 

Renewed emphasis on importance of multistakeholderism and crosscutting work on NETs

In line with the previous version of the resolution, this one included language on the importance of multistakeholderism and “ensuring the meaningful participation of all relevant stakeholders,” defined as including the private sector, academia, the media and civil society. Notably, this new version of the resolution lays out the specific value-add each stakeholder group can bring to helping address both the opportunities and challenges of rapid technological advancement. 

In a further endorsement of the cross-cutting impacts of NETs, the core group also added language on technical standard-setting and the “need to include a human-rights based perspective within standard-setting processes” for NETs. These additions are aligned with the OHCHR’s recent report on technical standard-setting processes for NETs—also submitted to the HRC at the 53rd session—and reflects the UN’s growing interest in thIs area.

Expanded role for OHCHR in developing guidance on NETs and human rights

The updated resolution tasks the OHCHR with several new responsibilities, including the development of a publicly accessible digital resource, the provision of advice and technical assistance to states. It will also continue its work on the practical application of the UN Guiding Principles on Business and Human Rights by producing an updated report on the challenges and good practices relating to the activities of technology companies, including by convening a multistakeholder consultation. 

This expanded OHCHR mandate could provide more opportunities for civil society engagement given the forum’s comparative openness and accessibility. However, without efforts to align the OHCHR’s work with that of other UN initiatives in the area of NETs, there is a risk of duplication of efforts in an already overcrowded field.

Next steps

Overall, the revisions to the NET resolution are positive. They both show that there are baseline areas of consensus among member states, and, more broadly, signal a growing interest across the UN system in AI and technical standards as they relate to human rights. 

We hope that the resolution can serve to improve and streamline coordination across UN agencies and bodies to ensure human rights remain at the centre of NET development, deployment, and usage, particularly with regards to AI. To do this, it’s critical that the OHCHR’s multistakeholder consultation on its upcoming report on UNGPs, NETs and human rights provide meaningful opportunities for civil society to intervene and share feedback. As the resolution only calls for a multistakeholder consultation for this report, GPD also strongly recommends that the OHCHR host an additional multistakeholder consultation for its upcoming work mapping UN work in the field of human rights and NETs, in addition to consulting with member states.