When AI hurts people, who’s to blame? Global experts grapple with accountability
When AI Hurts People, Who’s to Blame? Global Experts Address Accountability
When AI hurts people who s - When AI hurts people, who’s to blame? This pressing question has dominated discussions at the United Nations’ first AI governance summit, as global leaders seek clarity on assigning legal responsibility for AI-driven harms. On the second day of the event, specialists highlighted how the technology is increasingly tied to human rights violations, prompting urgent debates about accountability frameworks and ethical oversight.
AI-Linked Child Exploitation Surges in Developing Nations
At the heart of the debate is the growing concern over AI’s role in child exploitation, particularly in the Global South. Sonia Livingstone of the Independent International Scientific Panel on AI reported that in 11 developing countries, one child per classroom has been exposed to explicit deepfakes generated by AI. These digital manipulations are becoming more frequent, with reports to the U.S. CyberTip line and the National Center for Missing & Exploited Children (NCMEC) rising sharply. The technology’s capacity to amplify abuse has intensified calls for stronger regulatory measures.
“The evidence of AI-related violations against individuals and vulnerable groups is far more convincing than the benefits it offers,” Livingstone stated.
She emphasized that AI is not only a tool for progress but also a means of manipulation, as self-learning systems exploit personal data to target susceptible populations. This dual nature of AI—where it can both empower and endanger—has led to a consensus that universal governance is essential to protect global rights. The UN is now advocating for inclusive policies that address the technology’s risks across all regions.
Environmental Impact and Inequality in AI Deployment
The environmental footprint of AI has emerged as another critical issue, with experts stressing the need for equitable distribution of its benefits. During the first Global AI Dialogue in Geneva, alongside the AI for Good Summit, discussions focused on the disproportionate impact of energy-intensive data centers. UN Secretary-General António Guterres called for greater transparency in these systems, noting that their energy demands are escalating, yet their negative consequences often fall hardest on marginalized communities.
“We trust medicines, cars, and aircraft because they meet safety standards. AI should be no different,” said Volker Türk, UN High Commissioner for Human Rights.
Türk underscored the importance of embedding accountability and oversight into AI’s design, ensuring it aligns with human rights principles. Meanwhile, Sasha Luccioni of the Sustainable AI Group pointed out that while AI access is widespread, its environmental toll—such as increased energy use and emissions—is concentrated in regions with fewer resources to mitigate these effects. This disparity has sparked a push for more sustainable AI practices.
Luccioni added that the Global South is often left bearing the brunt of AI’s environmental impact, even as its economic gains from the technology diverge from the local consequences. This imbalance highlights the urgent need for policies that address the uneven distribution of AI’s benefits and risks.
Complexities of AI Oversight and Disinformation
Monitoring AI’s impact remains a formidable challenge, according to Jhalak Kakkar of the Centre for Communication Governance. She warned that current methods of assessment risk deepening global inequalities, as AI’s benefits are often harvested in one region while its harms are felt in another. “How do we address disparities when profits from AI are generated far from the communities it harms?” Kakkar questioned, emphasizing the necessity of transparent metrics.
“AI’s capacity to generate disinformation is alarming, especially when it targets vulnerable groups,” Kakkar noted.
Amal El Fallah Seghrouchni from Morocco highlighted the complexity of AI systems, particularly large language models that operate with 180 billion parameters. “Few legal or social experts can fully grasp the intricacies of these algorithms,” she said, stressing the need to “open up the black boxes of AI” to ensure accountability. The challenge of understanding AI’s mechanisms underscores the importance of interdisciplinary collaboration in governance.
Compounding these issues is the underrepresentation of women in AI development. Sima Bahous of UN Women shared findings that 25% of women human rights defenders, activists, and journalists face AI-assisted online violence, with 6% specifically targeted by deepfakes. “So much of this goes undocumented, unnoticed, and unreported,” Bahous remarked, drawing attention to the gendered impact of AI-driven harm.
“When AI hurts people, it’s not just about technology—it’s about who gets protected and who gets left behind,” Bahous concluded.
The UN’s latest report reveals that women make up just 30% of the global AI workforce, with 88% of leading researchers being male. This gender imbalance, experts argue, risks perpetuating existing inequalities in AI governance. By ensuring diverse representation, stakeholders can better address the social and ethical dimensions of AI’s impact.