The United Nations’ first Global Dialogue on AI Governance, convened in Geneva on 6–7 July 2026, opened with a preliminary assessment from its newly formed Independent International Scientific Panel on Artificial Intelligence that frames the central tension of the moment: AI capabilities are advancing faster than the institutions designed to govern them. The panel, co-chaired by Yoshua Bengio and Maria Ressa, flagged the technology’s capacity to generate persuasive deceptive content at scale, contributing to what it describes as a gradual erosion of information integrity. The gathering drew delegates from 108 countries, but the asymmetry it seeks to address is starkly material: the United States houses 75 per cent of the computing capacity of the world’s 500 most powerful AI supercomputers, while China holds 15 per cent, leaving the rest of the world with a sliver of the hardware on which frontier models depend.

Alongside the diplomatic track, a cluster of new labour-market and security data is giving shape to the uneven impact the panel warns of. Australia’s Department of Employment and Workplace Relations reports that AI has not yet triggered large-scale job losses—unemployment stood at 4.4 per cent in May 2026—but its first national mapping shows that the most-exposed occupations, including telemarketers, accountants and advertising professionals, are growing at 5.6 per cent, roughly half the pace of the least-exposed roles. Women and university graduates are disproportionately concentrated in those higher-exposure categories. An International Labour Organization study covering 11 ASEAN economies finds that while nearly 80 million workers have some exposure to generative AI, only 3.3 per cent—about 11.7 million people—fall into the highest-exposure bracket, and no evidence of mass displacement has yet appeared. Singapore registers the region’s highest share of exposed jobs at 42.2 per cent, and women across the bloc are more than twice as likely as men to work in high-exposure roles.

The security dimension is sharpening in parallel. In Mexico, deepfake-enabled identity fraud now accounts for 23.3 per cent of global attacks, according to a study by Unico and Liminal, with the cost of executing advanced impersonation attacks having fallen more than a hundredfold. Criminal networks reuse synthetic identities across dozens of firms, and nearly nine in ten fraud-prevention specialists view synthetic content as a principal threat. The UAE Cybersecurity Council separately links AI-powered phishing to more than 90 per cent of digital breaches, though research also indicates that almost half of targeted users still pause or report suspicious messages, suggesting that trained human judgment remains a meaningful control.

Governments are responding with markedly different levels of institutional readiness. The second edition of the Global Index on Responsible AI, covering 135 countries, records an average score of just 35 out of 100, with evidence of implementation present in only 55 per cent of cases where frameworks exist—a figure that drops to 45 per cent in the Global South. Indonesia used the Geneva platform to propose a global coalition on child protection in AI, citing its own PP TUNAS regulation as a model, while the World Bank Group is steering its efforts toward “small AI” applications that run on local devices and languages, targeting frontline health and education workers. The next concrete milestone is the final report of the scientific panel, expected to translate its preliminary findings into actionable recommendations, alongside a promised update to Australia’s national AI regulatory framework in the coming week.