A nuclear weapon is detonated in an attack

Where things stand (17 Oct 2026)

Nothing qualifying has happened since the question opened on 16 Sep 2026. No weapon has been used in war, and neither the CTBTO nor any of the five governments has said one was. Still, the background is about as tense as it has been since the Cold War:

• US–Israel war on Iran. It began on 28 Feb 2026. Khamenei was killed at the start, a June ceasefire collapsed, and the US is preparing renewed strikes after the 3 Nov midterms. Iran has no bomb. Hardliners are pushing for withdrawal from the Non-Proliferation Treaty and some talk openly of building a weapon. Trump explicitly ruled out US nuclear use on 31 Aug and again on 15–16 Oct.
• Russia. The Ukraine war continues; the think tank ISW reports a small net Russian territorial loss in September. On 30 Sep Russia sent NATO a diplomatic note threatening nuclear use if Kaliningrad were cut off, and Putin repeated it at the Valdai forum on 1 Oct. The last US–Russia arms treaty (New START) expired in February 2026.
• India–Pakistan. They fought a four-day air war in May 2025, their warships collided in Sept 2026, and the Indus waters dispute is still open.
• Korea. North Korea has made four missile launches in a month and there was a border mine blast.
• US–China. Relations look relatively calm after the September Trump–Xi summit.

Base rate and reference forecasts

• History: no nuclear weapon has been used in war in 81 years, including during defeats (Vietnam, Afghanistan, Russia's 2022 retreats) and direct clashes between nuclear powers (1969, 1999, 2019, 2025).
  • Laplace's rule gives about 1.2% a year; a Jeffreys-prior estimate gives about 0.6% a year.
  • A 1% annual rate is consistent with the record: it implies a 44% chance of 81 clean years. I computed this myself.
• Superforecasters: in a 2022 tournament they gave 4% by 2030 for a nuclear use killing more than 1,000 people. In a 2024 survey their median was 0.5% for a deliberate detonation by 2030; experts said 1% deliberate plus 1.3% inadvertent.
• Metaculus community, now, for any non-test detonation (accidents included): about 0.5% before 2027, 2.4% before 2028 and 5.1% before 2030.
• Prediction markets: Polymarket gives 3.3% for an Iranian bomb before 2027, and 24% for some NATO–Russia military clash by year-end (a low bar that includes shooting down a drone).

My model

Rough contributions over the window: Russia/NATO about 1.5%, the Iran conflict about 1%, India–Pakistan about 0.7%, US–China/Taiwan about 0.6%, Korea about 0.5%, terrorism or unauthorised use about 0.4%. That comes to roughly 1% a year.

Annual hazards I used:

• 1.3% for the rest of 2026 (17 Oct–31 Dec): Iran strikes likely resuming and the Kaliningrad standoff.
• 1.15% in 2027.
• 1.0% in 2028.
• 0.95% a year for 2029–2031.

If a real wartime detonation happened, a qualifying confirmation is very likely (I put it at 97%). Rivals would condemn it officially: the US over Russian, Iranian or North Korean use, Russia and China over Israeli use. I added 0.05% a year for edge cases: a wartime Iranian or Russian "test" being read as a compelling demonstration, or a contested official claim. Taking the cumulative product gives 0.27% by end-2026, 1.4% by end-2027, 3.4% by end-2029 and 5.25% by end-2031.

Reading adopted: tests at test sites count as excluded tests even if they happen while the country is at war.

Strongest objections

• For higher: several wars at once involving nuclear states, a formal Russian nuclear warning, arms control gone, a weakened Iranian regime, and Metaculus implying about 7% by 2031. I stay below Metaculus because that community has tended to overestimate nuclear use; its 2022 forecasts for 2023–24 resolved NO.
• For lower: superforecasters imply around 1–3%, and the taboo has held through 4.6 years of the Ukraine war. I weight the historical base rate, plus today's unusual number of hot conflicts, above those medians.

Sources

• Al Jazeera, "Trump rules out using a nuclear bomb on Iran" (16 Oct 2026): https://www.aljazeera.com/news/2026/10/16/trump-rules-out-using-a-nuclear-bomb-on-iran-why-that-matters
• Al Jazeera, "US–Iran stalemate" (15 Oct 2026): https://www.aljazeera.com/news/2026/10/15/us-iran-stalemate-what-are-trumps-next-military-options
• CFR, Froman (Oct 2026): https://www.cfr.org/articles/the-victory-of-asymmetry-in-the-iran-war
• Israel Hayom (10 Sep 2026): https://www.israelhayom.com/2026/09/10/fear-of-collapse-could-push-iran-toward-a-nuclear-bomb/
• Reuters via Internazionale, Putin at Valdai (1 Oct 2026): https://www.internazionale.it/ultime-notizie-reuters/2026/10/01/putin-says-the-world-is-living-through-a-dangerous-moment-warns-west-not-to-escalate
• Al Jazeera, Kaliningrad note (30 Sep 2026): https://www.aljazeera.com/news/2026/9/30/russia-warns-nato-of-possible-nuclear-response-if-kaliningrad-is-cut-off
• VOA on Putin's nuclear threats and New START's expiry: https://www.voanews.com/a/russia-putin-reiterates-nuclear-threat-against-nato/8208083.html
• ISW assessment (2 Oct 2026): https://www.criticalthreats.org/analysis/russian-offensive-campaign-assessment-october-2-2026-6ac053ceab44c
• Al Jazeera, India–Pakistan warship collision (17 Sep 2026): https://www.aljazeera.com/news/2026/9/17/india-pakistan-navy-ship-collision-sparks-alarm-over-unintended-escalation
• Forecasting Research Institute nuclear risk study (2024), Table 4: https://forecastingresearch.org/research/nuclear-risk
• 2022 forecasting tournament report, Q31: https://forecastingresearch.org/pdf/existential-risk-persuasion-tournament.pdf
• Metaculus Q42629 (search-result snippets only; the page itself was blocked)
• Polymarket API (gamma-api.polymarket.com), read 17 Oct 2026

Question. By the last day of each quarter, has a nuclear weapon been detonated in an attack or act of war on or after 2026-09-16, as confirmed by the CTBTO or by the government of any NPT nuclear-weapon state?

Definitions.

• Nuclear weapon detonated: an explosive device used as a weapon produces a nuclear explosion through fission and/or fusion. Any positive nuclear explosive yield counts, including a partial-yield detonation; no minimum yield, casualties, or damage is required.
• Attack or act of war: deliberate hostile use against people, territory, infrastructure, equipment, or military forces, or a deliberate wartime demonstration intended to compel an adversary. This includes terrorist and other non-state attacks, unauthorized hostile use, defensive use, and detonations in the atmosphere, space, underwater, or underground. Neither a declaration of war nor identification of the perpetrator is required. AI involvement is not required.
• NPT nuclear-weapon states: China, France, Russia, the United Kingdom, and the United States, fixed for this question regardless of subsequent treaty participation.
• CTBTO: the Preparatory Commission for the Comprehensive Nuclear-Test-Ban Treaty Organization, including its Provisional Technical Secretariat and Executive Secretary, or the treaty organization succeeding it upon entry into force.
• Confirmed: a qualifying source affirmatively states that a nuclear-weapon detonation occurred and that it was an attack or act of war, or describes facts unambiguously meeting those definitions. Statements from more than one qualifying source may jointly establish these elements for the same event. A statement of suspicion, possibility, or ongoing investigation is insufficient.

Resolves YES if the qualifying confirmation has been published by the quarter-end. Confirmation by one of the specified bodies is sufficient; agreement by the others is not required. A denial by the alleged perpetrator does not veto an otherwise qualifying confirmation.

Exclusions (resolve NO).

• Nuclear tests, exercises, peaceful nuclear explosions, and weapons demonstrations outside actual hostilities, even if described as provocative or unlawful.
• Accidental explosions, including accidents during a war, unless the qualifying source confirms deliberate hostile nuclear use.
• Dirty bombs, reactor damage or meltdowns, conventional explosions involving nuclear materials, and detonation of a warhead's chemical explosives without nuclear explosive yield.
• Launches, threats, intercepted or failed weapons that produce no nuclear explosion, and sensor detections that do not establish both nuclear-weapon detonation and hostile use.
• Historical references, including new publications describing Hiroshima or Nagasaki, and all detonations before 2026-09-16.
• Absence of a qualifying confirmation by the horizon resolves NO for that horizon, not a finding that no event occurred.

Resolving source. This question uses a narrower source rule than Public confirmation (A2): only CTBTO official statements or reports published at ct bto.org (the organization's domain, ctbto.org), or official statements, reports, and publicly archived briefing transcripts of the five named governments count. Government sources include the head of state or government, foreign or defence ministry, armed forces, intelligence agencies, and diplomatic missions, speaking in an official capacity and published in an official government or UN record. Media reports, unnamed officials, personal posts, legislative members' individual claims, and statements by other governments or international organizations do not count. The A2 developer-dispute rule does not apply to these expressly named resolving bodies. Apply the Resolution window and Resolution date (A3); where multiple documents are needed, use the publication date of the last document needed to establish all elements.