Why the UN’s first scientific report on AI wants governments to act before it’s too late
The UN's Independent International Scientific Panel on AI released its first preliminary report, billed as the first comprehensive global scientific assessment of artificial intelligence
The 40-member expert panel examined AI's impact on healthcare, education, the economy, security, human rights, and governance
The report warned that AI capabilities are advancing faster than governments' ability to understand, test, or regulate them, and flagged that computing power for the most advanced AI systems remains concentrated in very few countries
The report was released ahead of the first UN Global Dialogue on AI Governance, held in Geneva from 6–7 July 2026, where an Indian delegation participated
Independent International Scientific Panel on AI (IISPA)
The IISPA is the UN's first global scientific body dedicated entirely to artificial intelligence, modelled loosely on the IPCC's role for climate science. It was formally established through UN General Assembly Resolution A/RES/79/325, adopted by consensus on 26 August 2025, with its mandate rooted in the Global Digital Compact adopted under the Pact for the Future (Summit of the Future, September 2024).
Key Details
- Comprises 40 independent experts serving in personal capacity, with balanced geographic and gender representation, appointed for three-year terms
- The 40 members were appointed by the UN General Assembly on 12 February 2026; the panel elected Yoshua Bengio and Maria Ressa as co-chairs at its inaugural meeting on 3 March 2026
- Mandate is to issue one annual, policy-relevant but non-prescriptive scientific assessment report — it does not set binding rules, unlike a regulator
- Operates strictly in the non-military domain
The July 2026 report is the panel's first output under this mandate, intended to give governments an evidence base before the Global Dialogue on AI Governance negotiates cooperative approaches to regulation.
UN Global Dialogue on AI Governance
The Global Dialogue on AI Governance is a universal, multi-stakeholder UN forum — also created by Resolution A/RES/79/325 — where member states, industry, academia, and civil society deliberate on international approaches to AI. Its first session opened in Geneva on 6 July 2026.
Key Details
- Structured around four thematic clusters: AI's social and economic implications, bridging AI divides, safe and trustworthy AI, and human rights in the AI context
- Distinct from binding regulatory instruments — it is a deliberative/consultative mechanism, not a treaty-making body
- Receives the IISPA's annual scientific report as its evidentiary input
The Global Dialogue is the institutional venue where the panel's warnings about compute concentration and regulatory lag are meant to translate into coordinated government action.
Compute Concentration and the Global Digital Divide
The panel's central concern is the concentration of AI computing infrastructure: the United States accounts for roughly three-quarters of the computing power behind the world's most capable AI systems, while China holds close to 15%, leaving other regions — including most of the Global South — with minimal access to frontier-level compute.
Key Details
- This asymmetry underpins debates on "AI divides" — a recurring theme at the Global Dialogue on AI Governance
- Contrasts with the EU's approach: the EU AI Act (Regulation 2024/1689) entered into force on 1 August 2024 as a binding, risk-based framework (unacceptable/high/limited/minimal risk tiers), one of the few enforceable AI laws globally, unlike the UN's non-prescriptive scientific-assessment model
- India has pursued a distinct middle path: the IndiaAI Mission, approved by the Union Cabinet in March 2024 with an outlay of over ₹10,300 crore, focused on expanding affordable compute access (from an initial target of 10,000 GPUs to about 38,000 GPUs) rather than binding regulation
India's delegation at the Global Dialogue pushed the theme of "inclusive AI" — access to compute and AI benefits for developing countries — directly responding to the compute-concentration finding highlighted in the panel's report.
- IISPA established via UNGA Resolution A/RES/79/325, adopted by consensus on 26 August 2025
- Panel size: 40 members, 3-year terms; co-chairs elected 3 March 2026 (Yoshua Bengio, Maria Ressa)
- US share of top AI-training compute: ~75%; China: ~15% — together roughly 90% of frontier compute
- EU AI Act entered into force: 1 August 2024; full applicability for high-risk systems: 2 August 2026
- IndiaAI Mission outlay: ₹10,371.92 crore (approved March 2024), implemented by IndiaAI under MeitY
- Global Dialogue on AI Governance first session: Geneva, 6–7 July 2026