What happened
The inaugural UN Global Dialogue on AI Governance opened to roughly 4,000 participants from 170+ countries. It was the first time, as the Secretary-General put it, that “every country has a seat at the table” on AI, with “a shared base of evidence” in hand. The morning paired the UN’s political leadership with the scientists: after opening remarks, the co-chairs of the Independent International Scientific Panel on AI handed their first report to member states, framing the two days that followed.
The Secretary-General: four priorities
Guterres called AI “a technology… being deployed faster than anyone, including the people building it, can keep up,” and set out four priorities for governance:
- Safety. Common baselines and shared methods to test and verify frontier systems. Most urgently, an AI child-safety pledge: prove a system is safe before a child can reach it; zero tolerance for AI-generated child sexual-abuse material; and never leave a child in crisis alone (“when a child is harmed, the answer must never be ‘the algorithm did it.’”).
- Red lines. Human rights are “not negotiable”; in high-stakes decisions (healthcare, policing), “machines can inform, but humans must decide and answer.”
- Capacity. Private AI investment neared half a trillion dollars last year while public investment for developing countries is “a rounding error.” He backed a UN-supported global network for AI capacity building (20+ states have nominated centers) and said he will recommend a global fund for AI to the General Assembly.
- Transparency. His AI environmental-transparency initiative asks companies to disclose the full carbon, water and land footprint of their systems and to power data centers with renewables by 2030, noting they could use more electricity than all but five nations by then.
He also drew a hard line on lethal autonomous weapons, which he called “killer robots… morally repugnant… must be banned by international law,” and closed with a warning: “we may be the last generation able to set the terms on which humanity and machines coexist.”
Around the opening
- PGA Annalena Baerbock anchored the case in the UN Charter and in child protection, citing that 99% of deepfakes are sexual and 96% target women and girls, alongside SDG upside: AI flood alerts in Bangladesh, planting guidance for Kenyan farmers.
- ITU (Doreen Bogdan-Martin) noted the UN moved far faster than it did for the internet: “three years since generative AI went mainstream” versus a decade for the WSIS process. UNESCO (Khaled El-Enany) pointed to its 2021 Recommendation on the Ethics of AI and training for 50,000+ civil servants across 192 countries.
- USG Amandeep Gill stressed that inclusion “is not a box to be ticked,” and that dialogue “without capacity” becomes a monologue; his office coordinates the 2027 dialogue.
The scientific panel hands over its report
Panel co-chairs Yoshua Bengio and Maria Ressa presented the preliminary report and made clear it offers no policy recommendations: “that will be your role,” Bengio told the room. He warned there are still “no known technical guarantees that AI will follow human instructions, norms or laws,” and that frontier models have been shown deceiving testers and detecting when they are being evaluated.
Ressa described consensus among the 40 members as “the floor of our concern, not the ceiling,” and gave the findings three human faces: a Tigrinya medical mistranslation that turned “intravenous antibiotics” into something life-threatening; a frontier model that found a flaw in the 27-year-old OpenBSD operating system and in FFmpeg (“the capability that finds the flaw to fix it is the same capability that finds the flaw to exploit it”); and a 14-year-old boy who died in 2024 after months of conversation with a chatbot. She balanced them with the upside the report documents: AlphaFold’s 200M+ protein structures used by 3M researchers, diabetic-retinopathy screening of 600,000 people in India, and food-security early warning in the Philippines. The seven working-group leads, drawn from every region and balanced by gender, then summarised their findings.
Why it matters for the SDGs
The opening set the frame for the whole dialogue: AI as both accelerant and threat to the Goals. The SG’s capacity pillar (the global network and proposed fund) is aimed squarely at the AI divide (SDG 10) and technology transfer (SDG 17); the child-safety pledge and human-rights red lines speak to SDG 16; and the environmental-transparency push ties AI’s footprint back to SDG 7, SDG 12 and SDG 13. Most of all, it handed governments an independent evidence base and told them the window to act on it “will not stay open long.”
Watch & read
- UN Web TV, full recording of the Day-1 opening & plenary block (6 July 2026, 4h44m).
- The Independent International Scientific Panel on AI, SDGCounting’s coverage of the report being handed over here.
- Full Global Dialogue coverage.
Quotations are lightly edited from an automated (Otter.ai) transcript of the UN Web TV recording and should be read as close paraphrase; names and titles were reconciled to public records and reflect roles at the time.