What happened
The final cluster put AI under a human-rights lens, co-led by UN Human Rights and the World Bank. High Commissioner Volker Türk opened with the analogy that framed the room: a new drug takes 10 to 15 years of trials and review before it reaches patients, and many fail on safety grounds even then. “Are we taking the same care with artificial intelligence?” His argument was that the binding framework already exists, in international human rights law, and the task is to embed it into how AI is designed, deployed and used, rather than to invent principles from scratch.
Türk: rights as the compass
Türk rejected the idea that regulation is the enemy of innovation. Safety standards for medicines, cars and aircraft are “the reason people trust those technologies in the first place.” Human oversight, he warned, “cannot be just a rubber stamp”: it needs a named person with the authority, competence, time, independence and power to alter or stop a system, especially as AI moves into healthcare, hiring, credit, migration and military decisions, where responsibility can “disappear into the system.” He announced the launch of a Human Rights Data Exchange, described as the first open, authoritative service documenting where rights are under threat, and pointed to his office’s new guidance on protecting children online. Quoting Lord Acton (“power tends to corrupt”), he set the task plainly: ensure AI serves people, not the other way around.
The co-chairs: who is in the room, and who is not
Linda Bonyo of the Lawyers Hub spoke for the people absent from the auditorium: gig workers managed by algorithms in Kenya, younger generations, and civil-society and global-South participants who struggled with visas and cost to attend. Her organisation launched an Africa AI Governance Index tracking AI strategies, laws and institutions across every African state. “We see you,” she said, “and we will speak for you.” Spain’s Minister Óscar López framed AI as “a political tool, not a consumer good,” pointed to Spain’s AI-supervision agency (the first in Europe), its digital-rights charter and a bill to protect minors online, and drew the line that closed his remarks: no digital technology stands above a single human right.
The panel: accountability, gender, and children
Moderated by Canada’s Ambassador David Lametti (a former Justice Minister), the panel turned to how justice systems deliver accountability when a machine makes the decision. UN Women’s Executive Director Sima Bahous made the gender case with evidence from the scientific panel and her own agency:
- The numbers. 44% of assessed AI systems show gender bias; nearly one in four women human-rights defenders and journalists surveyed by UN Women had faced AI-assisted online violence; up to 99% of online deepfakes and manipulated sexual imagery target women; 41% of women self-censor online to avoid abuse.
- The gap is implementation, not principles. States remain the primary duty-bearers under existing law (the Universal Declaration, the ICCPR, CEDAW), and companies “cannot hide behind the complexity of their systems”; responsibility runs the whole AI life cycle.
- Children. Canada described an “AI for All” approach and a proposed rule that a social platform or chatbot must be proven safe before children can reach it, echoing the Secretary-General’s child-safety pledge from Day 1; scientific-panel member Sonia Livingstone anchored the child-rights evidence.
Why it matters for the SDGs
Human rights are the connective tissue between AI and the Goals. The accountability-and-justice thread maps to SDG 16 (rule of law, access to justice, accountable institutions); the gender evidence to SDG 5; and the “who is left out” framing to SDG 10. Türk’s reminder that “every country is involved in the AI value chain” (through data, minerals, labour, compute or cloud) and so should have a stake is the SDG 17 case for inclusion, and his call to steer AI investment toward health, education, social protection and climate ties directly to SDG 3, SDG 4 and SDG 13. The measurement angle SDGCounting watches: rights are only protected if harms can be documented, which is what the new Human Rights Data Exchange sets out to do.
Watch & read
- UN Web TV, recording of the Day-2 thematic sessions (7 July 2026).
- Independent International Scientific Panel on AI, the evidence base several speakers cited.
- Full Global Dialogue coverage.
Quotations are lightly edited from an automated (Otter.ai) transcript of the UN Web TV recording and should be read as close paraphrase; names and titles were reconciled to public records and reflect roles at the time.