UN AI summit under scrutiny
World leaders are debating artificial intelligence at the UN summit, but experts say summit diplomacy is necessary yet insufficient—deeper international cooperation and governance infrastructure are required. (businesstimes.com.sg) Policy voices at the same moment are pressing to make existential AGI risk a top priority and debating big solutions—from unified standards to the idea of an AI "constitution"—shifting the conversation from hype to hard governance choices. (80000hours.substack.com) (newyorker.com)
The UN General Assembly adopted Resolution A/RES/79/325 on August 26, 2025, creating an Independent International Scientific Panel on Artificial Intelligence and a Global Dialogue on AI Governance, with the Panel set to comprise 40 experts and to publish an annual, non‑prescriptive policy summary. (digitallibrary.un.org) India’s AI Impact Summit ran at Bharat Mandapam in New Delhi from February 16–20, 2026 and drew almost 300,000 participants from over 100 countries, where the New Delhi Frontier AI Impact Commitments secured voluntary pledges from Anthropic, Google and OpenAI to share anonymised economic data and strengthen multilingual evaluations. (businesstimes.com.sg) (impact.indiaai.gov.in) Anthropic CEO Dario Amodei told the summit that advanced AI could outstrip human cognitive performance in “only a small number of years,” and a GPAI assessment cited at the event warned that leading AI benchmarks could be saturated before 2030 with related security and control challenges likely to emerge by 2028. (businesstimes.com.sg) On February 20, 2026 the United States and India formalised cooperation on tech supply chains by signing the Pax Silica Declaration, a U.S.‑led initiative to secure critical minerals, semiconductors and AI infrastructure, and India also signed a joint India‑U.S. AI Opportunity Partnership at the summit sidelines. (state.gov) (mea.gov.in) Policy analysts and advocacy groups have pushed existential AGI risk to the top of the agenda, exemplified by an 80,000 Hours problem profile published February 24, 2026 arguing that advanced AI could transform civilisation within decades and that mitigating severe or existential risks should be a global priority. (80000hours.org) Anthropic published a public “Claude’s Constitution” on January 22, 2026—a 20‑ to 23,000‑word document led by philosopher Amanda Askell that the company says guides Claude’s values and training—while Jill Lepore’s March 30, 2026 New Yorker feature examined whether company‑level constitutions effectively shift responsibilities from democratic institutions to private firms. (anthropic.com) (newyorker.com)
