Summary: On February 4, 2026, Secretary-General António Guterres and Special Envoy on Technology Amandeep Gill put forward 40 nominees for the newly created Independent International Scientific Panel on Artificial Intelligence, handing the General Assembly the job of confirming them. Capital and model capability set new ceilings on a weekly basis; the UN’s bet is that what’s missing is something humbler — a common pool of facts the entire world can reason from.
The list is in. On February 4, 2026, Secretary-General António Guterres and Special Envoy on Technology Amandeep Gill put forward 40 nominees for the newly created Independent International Scientific Panel on Artificial Intelligence, handing the General Assembly the job of confirming them. In Guterres’s telling, the world’s first fully independent scientific institution devoted to closing the AI knowledge gap now has a founding roster.
Why this moment? By Guterres’s own description, AI is “moving at the speed of light.” That very day, markets were still working through word that Amazon is in talks to invest as much as $50 billion in OpenAI. Capital and model capability set new ceilings on a weekly basis; the UN’s bet is that what’s missing is something humbler — a common pool of facts the entire world can reason from.
Inside the Slate of 40
Nineteen women and 21 men make up the list, chosen for recognized AI expertise, with geographic spread and gender balance written into the selection criteria. UN News singled out three:
- Sonia Livingstone (United Kingdom): a professor at the London School of Economics who has spent years researching media literacy and children’s digital rights
- Balaraman Ravindran (India): who leads the Data Science and AI program at IIT Madras
- Maria Ressa (Philippines): a journalist who co-founded Rappler and holds a Nobel Peace Prize
That a working journalist and a children’s-rights scholar sit on a panel about “AI science” is the design, not an accident. Grading model capability is not the assignment — the job is to assess what AI actually does to societies.
AI’s Version of the IPCC
The blueprint is borrowed from climate science. Like the IPCC, the panel would belong to no bloc and would periodically condense evidence into assessment reports. Three duties define the mandate:
- Weigh the real effects AI is having on economies and societies
- Go deep on priority sectors: health, energy, and education
- Spread leading-edge research and narrow the knowledge gaps separating countries
Guterres described the purpose without decoration: helping the world “separate fact from fakes, and science from slop.” In a year when generative AI has made fabrication nearly free, the phrase reads as policy rather than poetry. Common understanding, he added, is what allows countries “to build effective guardrails, unlock innovation for the common good, and foster cooperation.”
Three Years from Advisory Report to Nominee List
Almost three years separate the idea from this slate. 2023 was the year ChatGPT turned AI into a universal concern, prompting Guterres to convene a High-Level Advisory Body on AI — the body whose “Governing AI for Humanity” report recommended creating exactly this panel. The nominations now sit with the General Assembly, due to finalize the 40 members on February 12, 2026. Terms last three years, and the opening report is expected before July.
Consensus Under Geopolitical Strain
The skeptical question writes itself: with powers squared off and multilateral machinery stalling, what can the UN plausibly deliver? Gill conceded that consensus-building is difficult amid today’s geopolitical tensions. His reply leans on history: scientists kept cooperating even at the height of the Cold War, and what the UN supplies are “mechanisms where scientific understanding, common understanding can be advanced.” The slate bakes in geographic representation, with coverage intended across the five UN regional groups, so the Global South shares the same evidence table where the wealthy AI-producing states are already seated.
That detail is the quiet heart of the exercise. Most governments operate no frontier labs of their own and cannot afford to audit vendor claims; a standing, independent synthesis hands them material to legislate from that traces to neither lobbying nor marketing.
The Read for Developers and Product Teams
Three consequences are worth planning around. First, the compliance map gains a “scientific consensus” tier: once the panel settles on a position about a given risk, national legislation tends to move in behind it — the path the IPCC cut through climate regulation. Second, the deep dives into health, energy, and education will amount to a ready-made evidence library for teams scoping AI products, and a more trustworthy one than the white papers vendors scatter around. Third, the assignment of telling “fact from fakes” leads straight to information integrity, so content platforms and generative AI products should be reserving engineering room for provenance and verification features right away.
Governance either matches “the speed of light” or it doesn’t, and two dates will say which: the February 12 confirmation, then the first report in July.