Nuclear Weapons Proliferation: The Threat We Stopped Watching

The Nuclear Threat That Never Disappeared

In January 2023, the Bulletin of Atomic Scientists moved its Doomsday Clock to 90 seconds to midnight — the closest it has ever stood to symbolic global catastrophe since the clock was established in 1947. That number has not improved since. It is not a metaphor for anxiety. It is a calibrated judgment by physicists, arms-control specialists, and former heads of state who examine the actual architecture of nuclear risk: weapons stockpiles, delivery systems, diplomatic breakdowns, and the accelerating erosion of the frameworks designed to keep these forces in check. Ninety seconds. We once stood at seventeen minutes. The distance between those two figures is not progress. It is the accumulated weight of three decades of inattention, and it is closing.

The world did not disarm after the Cold War ended. It renegotiated the terms of its danger. Nine states now possess nuclear weapons. Approximately 12,500 warheads remain in global arsenals, the overwhelming majority held by the United States and Russia. The rest — China, France, Britain, Pakistan, India, Israel, and North Korea — together account for several hundred more. The optimism of the 1990s, when START treaties and the indefinite extension of the Nuclear Non-Proliferation Treaty seemed to chart a course toward a smaller, safer world, has curdled. What replaced it was something more insidious than open hostility: drift, distraction, and the comfortable illusion that nuclear weapons proliferation was a problem the previous generation had more or less solved.

It was not solved. We simply stopped looking.

Accelerating Proliferation and the New Nuclear Landscape

The deterioration of arms-control architecture has been systematic and largely uncontested. The United States withdrew from the Intermediate-Range Nuclear Forces Treaty in 2019, citing Russian violations. New START, the last major bilateral treaty capping deployed strategic warheads, lapsed in early 2026 without a successor. Russia had already suspended its participation. The Open Skies Treaty, a verification instrument of genuine practical value, is now a historical artifact. Each of these collapses happened incrementally, reported briefly, and then absorbed into the background noise of other crises.

Scholars at the Carnegie Endowment for International Peace have documented what they call the "hollowing out" of the nonproliferation regime — a system that was never perfect but that at least provided shared norms, inspection regimes, and diplomatic forums where tensions could be managed below the threshold of catastrophe. The NPT Review Conference in 2022 ended without consensus, the fifth successive failure to produce a final document. The UN Treaty on the Prohibition of Nuclear Weapons, which entered into force in 2021 and has attracted signatures from more than ninety states, has been dismissed by every nuclear-armed state as aspirational at best and counterproductive at worst. The gap between those who want to abolish nuclear weapons and those who actually possess them has never been wider or more structurally entrenched.

China is modernizing and expanding its arsenal at a pace that analysts at the Arms Control Association describe as unprecedented in the post-Cold War era, moving toward a nuclear triad and potentially tripling its warhead count by the end of the decade. North Korea has conducted a series of missile tests demonstrating intercontinental reach. Pakistan and India continue to develop tactical nuclear weapons — short-range systems that lower the threshold for battlefield use. The nuclear landscape is not frozen. It is proliferating, quietly, beneath the headlines.

The Role of AI in Amplifying Nuclear Danger

The intersection of artificial intelligence and nuclear weapons represents a qualitatively new category of risk, one that arms-control frameworks built in the twentieth century were never designed to address. AI systems can accelerate weapons design, compress development timelines for delivery systems, automate targeting decisions, and — perhaps most dangerously — compress the decision window during a crisis in ways that leave less room for human judgment and miscommunication to be corrected before they become irreversible.

The danger is not merely theoretical. AI-assisted simulation can allow states to model nuclear weapon performance without conducting physical tests, enabling sophisticated warhead development that evades the traditional warning signals that the nonproliferation community depends upon. Autonomous systems capable of identifying and tracking mobile launchers could destabilize the second-strike calculations that underpin deterrence theory. A world in which both sides fear their retaliatory capability can be destroyed before it is launched is a world permanently on hair-trigger alert.

Against this backdrop, political leadership in the United States has not risen to meet the challenge. Donald Trump's dismissals of AI-related dangers — offered with the same breezy confidence that has characterized his approach to complex systemic risks — amount to a withdrawal from the intellectual seriousness the moment demands. Nuclear weapons and artificial intelligence are converging on a timeline that will not wait for the political calendar. The absence of serious high-level engagement with this convergence is not neutral. It is itself a form of risk accumulation.

Political Denial and the Culture of Scoffing

There is something historically familiar about the posture of dismissal. Every generation has produced its skeptics — those who argued that deterrence was stable, that rational actors would never cross certain thresholds, that alarmism was the greater danger. Sometimes they were right. The Cold War ended without nuclear exchange, and that fact is worth acknowledging honestly. But the intellectual habit of treating skepticism as inherently sophisticated and concern as inherently naive is a failure of epistemic hygiene, not a display of it.

The complacency that followed the Cold War was not irrational given the evidence of the 1990s. It became irrational as the evidence changed. When treaties collapse and are not replaced, when arsenals modernize without constraint, when new actors acquire capabilities and old actors expand theirs, the appropriate response is not continued calm. The appropriate response is updated assessment. What we have instead, in too many corridors of power and too many opinion pages, is the comfortable continuation of a prior conclusion that no longer matches the facts.

Political denial has costs. It deprioritizes funding for verification and monitoring. It allows expertise to atrophy in the agencies responsible for arms control. It signals to other states that the dominant powers are not serious about the rules they claim to uphold. These costs compound quietly over time and express themselves suddenly, in crises that arrive without warning.

Religious Prophecy, Secular Skepticism, and Real-World Risk

Across human history, the anticipation of catastrophic endings has never been far from the surface of religious and cultural life. The Book of Revelation describes a final period of tribulation — war, famine, plague, environmental collapse. Islamic tradition contains its own eschatological framework, centered on a day of judgment and reckoning. Christian evangelical theology includes the concept of a rapture, in which the righteous are preserved while the world convulses. These traditions vary enormously in their specifics, but they share a structural intuition: that human civilization is not permanently safe from civilizational rupture.

The secular response, at its best, is to take that intuition seriously while subjecting it to empirical scrutiny. At its worst, the secular response mirrors exactly what Matthew's gospel predicted: scoffing, dismissal, the comfortable assumption that those who warn of imminent catastrophe are either credulous or performing. The Bulletin of Atomic Scientists is neither. The Arms Control Association is neither. The scholars and former officials who move that clock forward are not apocalypticists. They are professionals who have spent careers studying the mechanisms by which nuclear weapons might be used, and they are telling us those mechanisms are under more stress than at any point since the Cuban Missile Crisis.

Skepticism is a virtue. Lazy skepticism — the kind that treats all warnings as equivalent regardless of the evidence behind them — is a liability.

Why We Must Look Again Before It Is Too Late

The case for renewed attention to nuclear weapons proliferation is not a case for despair. It is a case for seriousness. The structures that reduced nuclear risk in the twentieth century — arms-control treaties, verification regimes, diplomatic channels, shared norms — were built through sustained political effort over decades. They can be rebuilt. New frameworks capable of addressing AI-assisted weapons development, hypersonic delivery systems, and the expanded roster of nuclear-armed states are conceptually possible. They require political will, and political will requires a public and a political class that understands what is at stake.

The Doomsday Clock stands at 90 seconds to midnight not because catastrophe is inevitable but because the systems designed to prevent it are degraded and the political attention required to repair them is largely absent. That is a correctable condition. Correcting it begins with looking — honestly, persistently, without the comfort of assuming that because the worst has not yet happened, it cannot.

The bomb never went away. The architecture that contained it is crumbling. And in the place where serious policy engagement should be, there is, too often, silence punctuated by dismissal. That silence is not safety. It is its own kind of countdown.

Source: Opinion | The Guardian