UN Secretary-General Antonio Guterres warned that artificial intelligence is developing faster than the world's ability to govern it, and called for binding international rules to protect children from the technology's harms. The intervention lands as AI systems from OpenAI, Anthropic, Google, and Meta reach hundreds of millions of users, most of them without any regulator having reviewed the underlying models.
Guterres framed child safety as the most urgent gap. Generative AI tools capable of producing realistic images, synthetic voices, and personalized chat are already in the hands of minors through consumer products, and the guardrails vary widely by vendor. The Secretary-General wants a coordinated global response rather than the current patchwork of national rules.
That patchwork is the crux of the problem. The EU AI Act entered force in 2024 with staggered obligations running through 2027. The United States operates through a mix of executive orders, state-level bills, and sectoral enforcement by agencies including the FTC. China has its own generative AI rules. Most of the Global South has none. A model trained in one jurisdiction is deployed in all of them within hours.
Key facts
- 01UN Secretary-General Antonio Guterres warned that AI development is outpacing global oversight mechanisms.
- 02Guterres called specifically for international rules aimed at protecting children from AI-related harms.
- 03The intervention adds to a growing list of UN calls for coordinated AI governance across member states.
- 04No binding global AI treaty currently exists, leaving regulation fragmented across the EU AI Act, US executive orders, and national frameworks.
Guterres has been consistent on this point. He has repeatedly called for a UN-level AI oversight body modeled loosely on the International Atomic Energy Agency, and the General Assembly adopted a non-binding AI resolution in 2024. What is missing is enforcement — no treaty currently obligates a model provider to submit systems for review before shipping to children in any given country.
The child-safety framing is a deliberate choice. It is one of the few areas where governments across ideological lines tend to converge, and it sidesteps the harder political fights over speech, copyright, and labor displacement. The UK, EU, and US have each moved separately on AI-generated child sexual abuse material, deepfake harassment in schools, and age-verification requirements for chatbots. Guterres is pushing to consolidate those threads.
Industry response to prior UN AI statements has been muted. Frontier labs generally support the idea of international coordination in principle but oppose specific obligations that would slow shipping cadence or force disclosure of training data. Anthropic, OpenAI, and Google have all published their own child-safety commitments, though none are externally audited on a standardized basis.
The practical mechanism Guterres is proposing remains vague. A treaty would take years to negotiate and would require ratification by the major model-producing states — the US, China, France, the UK, and increasingly the UAE and Saudi Arabia. Absent that, a voluntary code of conduct is the more realistic near-term output, similar to the Hiroshima AI Process launched by the G7 in 2023.
Skeptics of UN-led AI governance argue the body moves too slowly to regulate a technology that changes on a quarterly release cycle. By the time a working group drafts principles, the frontier models have shifted underneath them. That criticism is fair, but the counter-argument is that no other forum has the legitimacy to bind non-Western AI developers, and unilateral Western rules will not reach the models being trained in Shenzhen or Abu Dhabi.
For AI companies, Guterres's intervention matters less as immediate regulatory risk and more as a signal of where the political center of gravity is heading. Child safety is the wedge issue that turns diffuse public unease about AI into specific legislative demands. Expect the EU, UK, and individual US states to cite the UN framing when they move on age verification, synthetic-media labeling, and chatbot design standards over the next 12 to 18 months — and expect the compliance cost to fall hardest on consumer-facing products rather than enterprise APIs.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
