UN AI Summit Highlights Rights Violations, Governance Gaps

The United Nations convened its first-ever summit on AI governance in Geneva this week, where international experts presented troubling evidence of artificial intelligence causing measurable harm to vulnerable populations worldwide.

Sonia Livingstone from the Independent International Global Scientific Panel on AI delivered findings from the panel's preliminary report, mandated by the UN General Assembly. The data showed that across 11 Global South countries, up to one child per classroom reported being targeted by sexually explicit AI-generated deepfakes. Reports of AI-created child sexual abuse material to the US National Center for Missing & Exploited Children's CyberTip line are climbing exponentially, according to details first reported by UN News.

Livingstone argued that evidence of AI-linked violations against individuals and disadvantaged groups now outweighs documented benefits. Self-learning systems are being deployed to amplify disinformation and manipulate personal data without adequate safeguards.

Why it matters

As AI capabilities expand faster than regulatory frameworks can adapt, the question of legal accountability remains unresolved. Without clear governance structures that span borders, vulnerable populations bear disproportionate risks while benefits concentrate in technologically advanced regions. The UN's two-summit process aims to establish universal standards before the governance gap widens further.

Energy and inequality concerns

Beyond direct harms to individuals, experts raised alarms about AI's environmental footprint. Sasha Luccioni, co-founder of the Sustainable AI Group, noted that energy-intensive data centers powering AI platforms create localized negative impacts far from where benefits accrue. Marginalized communities face worsening health effects from water consumption, energy demands, and emissions tied to AI infrastructure.

Jhalak Kakkar from Delhi's National Law University highlighted how this dynamic reinforces existing inequalities, with economic value flowing to distant locations while nearby communities absorb environmental costs.

The transparency problem

Morocco's Amal El Fallah Seghrouchni emphasized the technical opacity challenge: large language models operate continuously using approximately 180 billion parameters, creating "black boxes" that even AI researchers struggle to interpret. This complexity makes it nearly impossible for legal and social science experts to assess algorithmic decision-making or assign responsibility when systems cause harm.

UN Women Executive Director Sima Bahous presented survey data showing one in four women human rights defenders, activists, and journalists have experienced AI-assisted online violence. Six percent reported being victims of deepfakes or manipulated imagery. Much of this abuse goes undocumented, she noted.

The gender imbalance in AI development compounds these problems: women comprise only 30 percent of the global AI workforce, and 88 percent of leading AI researchers are male, according to the scientific panel's report.

Building accountability

UN High Commissioner for Human Rights Volker Türk called for embedding equality, accountability, and human oversight into AI systems from the design phase forward. He compared the need for AI safety standards to established protocols for medicines and aircraft—frameworks that build public trust rather than impede progress.

The Geneva dialogue represents the first of two summits, with a second session planned for May 2027 in New York. The UN International Telecommunication Union is hosting the proceedings to ensure participation from all nations, not just the most technologically advanced.

These details were first reported by UN News from the Global AI Dialogue in Geneva.