This is a policy argument from Norway's foreign affairs establishment, and it deserves to be read as such — not as neutral analysis but as a specific state's strategic bet that multilateral institutions are a national security asset, not a concession. The argument is structurally sound even if the messenger has obvious interests. The core thesis: we are entering genuine multipolarity (no single hegemon) at the exact moment AI is democratising the capacity for catastrophic harm. Cheap drones neutralise expensive militaries. Lab-engineered pathogens don't respect borders. Cyberattacks launch from garages. The US can't force Iran to comply; Russia can't subjugate Ukraine. Raw power no longer converts reliably into control. This is the diagnosis, and it is empirically defensible. The prescription is international institutions with real teeth — monitoring, inspection, information-sharing, enforceable global standards. The author cites the IAEA's verification regime as proof of concept, the WHO's Pandemic Agreement, the BBNJ high-seas biodiversity treaty, and a September 2024 Security Council debate on AI governance where most governments and AI developers called for stronger oversight. Norway and Finland launched a joint Call for Action on global oversight of frontier AI models. The risk math deserves attention: a 1 percent annual probability of global catastrophe translates to roughly 63 percent over a century. This is basic compound probability, not alarmism. It reframes the question from "is any single year dangerous?" to "can we sustain this risk level indefinitely?" The answer is obviously no, and the article is right to force that framing. The weakness is the gap between diagnosis and mechanism. The article identifies the problem with precision — no hegemon, proliferating threats, institutions underfunded — but the proposed solutions (UN reform, cross-regional coalitions, Norway's financial leverage as a major UN contributor) are structurally inadequate to the threat described. If AI-enabled bioweapons and autonomous cyberattacks are the threat, voluntary coalitions of like-minded democracies are a bandage on an arterial wound. The historical analogy is the article's sharpest and most uncomfortable move: effective international institutions emerged only after the catastrophes of World War I and II. The author explicitly asks whether we can build them this time without first experiencing another global catastrophe of comparable scale. He is honest enough to leave the question open, which suggests he suspects the answer is no. What Norway is actually doing is positioning itself as a broker in a fragmenting order — too small to impose outcomes, large enough (financially) to shape agendas. This is rational small-state strategy. The question is whether any number of like-minded coalitions can build enforcement mechanisms that great powers will accept when their interests diverge. History says they cannot, until the cost of not cooperating becomes unbearable. The article knows this. It says so in the final paragraph.
