The State of the Taboo—Non-Use, Threats, Tests, Treaties, and the Measurements

The Nuclear Taboo: Eighty-One Years of Non-Use, the Threats and Tests of 2025–2026, and the Treaties That No Longer Bind — Special Report

The last nuclear weapon used in war fell on Nagasaki on August 9, 1945. In the eighty-one years since, nuclear-armed states have lost wars, been attacked, threatened other states, and been threatened in return, and none has used one. This report files what that prohibition—named the nuclear taboo by the scholar Nina Tannenwald—consists of in November 2026: the behavior, which is undisputed; the measurements of public and expert opinion, which disagree with each other by instrument; the threats and the tests of 2025 and 2026, each graded as said, claimed, or done; the two wars fought over Iran’s program in which no side used a nuclear weapon; the treaties that have lapsed, ending with New START on February 5, 2026; and the public argument about use that ran through the autumn of 2026. Where the record does not agree with itself, the report prints the disagreement and does not average it.

By Eunice, the desk · Collected and analyzed at 0742 ET, Tuesday, November 10, 2026.

The File in Brief

The behavior is not in dispute; everything else about the taboo is. No state has used a nuclear weapon in war since August 9, 1945, and that record includes nuclear states fighting costly and losing wars against non-nuclear adversaries—the United States in Vietnam, the Soviet Union in Afghanistan. What holds the record in place is the contested part. Nina Tannenwald, whose 1999 article and 2007 book gave the prohibition its scholarly name, argues a normative taboo: a de facto prohibition against the first use of nuclear weapons, built by stigma. Her critics answer with weaker instruments—a tradition of non-use that rests on effects and reputation and can be altered, in T. V. Paul’s formulation, if material and political circumstances compel; a public aversion that survey experiments find to be weak and conditional; a norm its own defenders describe as under pressure. No researcher the desk found argues, on measurement, that the taboo is strengthening. The strongest claims for its health are behavioral: the weapons remain unused, including through the two most dangerous years of the record since 1962, and the cause of that non-use is unattributed.

The measurements divide by instrument, and the division is itself the finding. Asked in the abstract, American majorities say nuclear weapons should be used only in response to a nuclear attack, or never: 56 percent only-in-response and 21 percent never in a YouGov survey of October 30 to November 2, 2025, with 7 percent accepting first use in certain circumstances. Asked in concrete scenarios, majorities approve: in the best-known experiment, Sagan and Valentino’s 2015 Iran scenario, 59.3 percent approved a strike killing 100,000 Iranian civilians when the alternative was a ground invasion costing 20,000 American lives. Expectation runs an order of magnitude above approval in the same instruments: 46 percent of Americans in that same 2025 YouGov survey said nuclear war within ten years is likely, against the 7 percent who accept first use. The desk prints both families of numbers throughout this report, with their instruments named, and does not reconcile them, because the literature does not.

The events of 2025 and 2026 tested the prohibition at every level except use. On October 29, 2025, the President of the United States instructed his government, in a public post, to start testing nuclear weapons “on an equal basis,” and said on television three days later that the United States would test “like other countries do.” No American explosive test followed; his own Secretary of Energy redefined the tests as non-explosive within four days, and the verified negative stands at November 2026: the last American test was September 23, 1992, and no Russian or Chinese explosive test followed the exchange either. Russia revised its nuclear doctrine in November 2024, lowering the written threshold for conventional aggression and naming a new trigger; it fought in Ukraine for a fourth year under continuous nuclear signaling without use, and its warheads in Belarus remain claimed by two governments and unverified by every independent instrument. Israel and the United States struck Iran’s nuclear program twice, in June 2025 and again on February 28, 2026, the second time killing the Supreme Leader; Iran did not withdraw from the Non-Proliferation Treaty, and the International Atomic Energy Agency’s last verified count of its enriched uranium is dated June 13, 2025—440.9 kilograms at 60 percent—repeated unchanged in every report since, because the inspectors have not been readmitted. On February 5, 2026, New START expired by its terms, and for the first time since 1972 no legally binding limit governs the American and Russian strategic arsenals. In October and November 2026, the largest American broadcasters of the argument that a nuclear strike on Iran was under active consideration carried that claim to audiences in the millions; the claim is unverified, the President had denied it on the record on October 15, and the date one broadcaster set for the decision, November 7, passed with no strike documented in any source the desk retrieved. The taboo’s state, on the whole record, is this: unbroken in behavior for eighty-one years, unwitnessed by any strengthening measurement, and surrounded by the thinnest treaty architecture since the early 1970s.

1. The Baseline: A Weapon Unused in War Since 1945

The whole file rests on two dates. On August 6, 1945, the United States dropped a uranium bomb on Hiroshima; on August 9, 1945, it dropped a plutonium bomb on Nagasaki. Those remain the only wartime uses of nuclear weapons by any state. In the eighty-one years since, nine states have acquired the weapons, and nuclear-armed states have fought long and losing wars without using them: the United States lost more than 58,000 dead in Vietnam and the Soviet Union about 13,400 in Afghanistan, in the count Press, Sagan, and Valentino published in 2013, and neither reached for the arsenal. Non-use as behavior is undisputed by every camp in the literature. Its cause—norm, deterrence, military uselessness for the purpose at hand, or precedent—is disputed by all of them, and the dispute is the subject of this report.

The prohibition was tested first in Korea, and the record there is more complicated than its usual compression. The Joint Chiefs considered nuclear use as early as July 9, 1950, when General Bradley raised it and found no support; B-29 bombers deployed to Guam that month carrying bombs without their fissile cores. On November 30, 1950, President Truman told a press conference that use of the atomic bomb “always has been” under active consideration, that “it is one of our weapons,” and that the field commander would have charge of the weapons’ use; the White House clarified the same day that only the President could authorize use, and that no authorization had been given. General MacArthur requested commander’s discretion on December 9, 1950, and filed a target list on December 24—the bomb figures in his requests conflict by source, at 26, 34, and 38, and the desk does not resolve them. Truman relieved MacArthur on April 11, 1951, officially for statements contradicting administration policy; the historian Bruce Cumings’s thesis that nuclear reliability drove the relief is an attributed interpretation, filed as such. On April 5 and 6, 1951, the Joint Chiefs drafted conditional retaliation orders and Truman arranged the transfer of nine Mark 4 bombs to military custody. No use occurred. Joseph Nye, writing at Harvard’s Belfer Center on August 6, 2020, put the founding this way: Truman “created what Tom Schelling called the all important ‘nuclear taboo’ when he refused General MacArthur’s request.”

The survivors’ organization holds the prohibition’s institutional memory, and the Nobel Committee said so in the taboo’s own word. Nihon Hidankyo, the Japanese confederation of atomic-bomb survivors founded in 1956, received the Nobel Peace Prize for 2024 “for its efforts to achieve a world free of nuclear weapons and for demonstrating through witness testimony that nuclear weapons must never be used again.” The Committee’s press release states that the hibakusha “contributed greatly to the establishment of the nuclear taboo,” and continues: “It is therefore alarming that today this taboo against the use of nuclear weapons is under pressure.” The Committee delivered that sentence on October 11, 2024. The two years since are the balance of this file.

2. The Idea: Tannenwald, Schelling, and the Critics

The name is Nina Tannenwald’s, and its publication record is exact. Her article “The Nuclear Taboo: The United States and the Normative Basis of Nuclear Non-Use” appeared in International Organization 53(3) in the summer of 1999, pages 433 to 468. A second article, “Stigmatizing the Bomb: Origins of the Nuclear Taboo,” followed in International Security in the spring of 2005; the 2005 formulation defines the taboo as “a de facto prohibition against the first use of nuclear weapons.” The book, The Nuclear Taboo: The United States and the Non-Use of Nuclear Weapons Since 1945, was published by Cambridge University Press in December 2007—not 2005, as it is often misdated—and won the 2009 Joseph Lepgold Prize. Its cases are Japan 1945, Korea, Vietnam, and the Gulf War of 1991. The argument, in her construction, is that non-use is not explained by deterrence or by the weapons’ impracticality alone: the weapons became stigmatized, set apart from conventional arms as a category it is not done to use, and the stigma does causal work.

The idea has an older patent. Thomas Schelling, in his Nobel lecture of December 8, 2005, credited a taboo that Secretary of State John Foster Dulles perceived as early as 1953, and called the resulting convention an asset “not guaranteed to survive.” The lecture is twenty years old at this filing. Its caution is the caution every later assessment in this file repeats in its own vocabulary.

The critics are named, and their instruments are weaker by design. T. V. Paul, in The Tradition of Non-Use of Nuclear Weapons (Stanford, 2009) and in “Taboo or Tradition?” (Review of International Studies, 2010), argues for a tradition rather than a taboo: a practice resting on the weapons’ effects and on reputation costs, weaker than a prohibition, and one that “can be altered if material and political circumstances compel.” Rebecca Hersman Gibbons and Keir Lieber, in “How Durable Is the Nuclear Weapons Taboo?” (Journal of Strategic Studies, 2019), conclude the norm is “much more fragile than most analysts understand.” Daryl Press, Scott Sagan, and Benjamin Valentino, in the survey work filed in the next section, find in the American public “only a weak aversion,” with few of a taboo’s characteristics. Stacie Goddard and her co-authors, in “Nuclear Shibboleths” (International Organization, Winter 2025), argue the rationalist-versus-normative argument itself obscures the distinct pathways by which states arrive at non-use. Reid Pauly, from wargame evidence in International Security in 2018, counts five logics of restraint—deterrence, practicality, precedent, reputation, and ethics—operating together. The desk notes one absence: the phrase “paper taboo,” which circulates as if it were a named scholarly critique, was not found as a stated position in any work the desk retrieved. The documented equivalents are Paul’s tradition, Press and his colleagues’ weak aversion, and Gibbons and Lieber’s fragility, and the report uses those.

Tannenwald’s own later assessment is the measured center of the file. Writing in The Washington Quarterly and in Foreign Affairs in 2018, in an essay titled “The Vanishing Nuclear Taboo?”—the question mark is hers—she held that the taboo “remains a widely-shared norm” that is “under pressure,” with the disarmament order unravelling around it. The desks of the world have added eight years of record since. What they have not added, in any source the desk found, is a measurement showing the norm stronger than it was.

3. The Measurements: Two Survey Families That Do Not Agree

The central measurement fact of this report is a divergence, and it is replicated. Surveys that ask about nuclear use as an abstract principle return one public; experiments that present a concrete scenario return another. Both series are real measurements of real respondents. Which of them measures the taboo is contested in the literature, and no bridge between them has been established. The desk prints both, with instruments and dates, in that order.

The abstract-principle family is the older instrument. The wording family used by the Chicago Council on Global Affairs and its predecessors asks whether nuclear weapons should be used only in response to a nuclear attack, never, or in a first strike in certain circumstances. In 2004 (PIPA and Knowledge Networks, 1,195 respondents), 57 percent said only in response, 22 percent never, and 19 percent accepted first use in certain circumstances. In 2007 (PIPA and CISSM, 1,247 respondents), the only-in-response share split 50 percent among Republicans and 57 percent among Democrats, and the survey’s authors note that across the 2002, 2004, and 2007 administrations of the question, the first-use share “has never risen above one in four.” The 2010 Chicago Council administration, which the desk holds only by secondary citation, returned 57 percent and 20 percent on the same two answers. The family’s most recent administration the desk retrieved is YouGov’s, October 30 to November 2, 2025, 1,148 American adult citizens: 56 percent only in response, 21 percent never, 7 percent first use in certain circumstances. Across twenty-one years and four instruments, the abstract public does not move much: a majority for retaliation only, a fifth for never, and a first-use minority in single digits or the teens.

The scenario family returns a different country. Press, Sagan, and Valentino, in “Atomic Aversion” (American Political Science Review 107(1), February 2013, pages 188 to 206—the venue is the American Political Science Review, not International Security, and the third author is Valentino), embedded the nuclear option in a concrete choice and found preference for it rising from 18.9 percent, in the cell where a conventional strike was equally effective, to 51.4 percent where the nuclear option was more effective. Sagan and Valentino’s “Revisiting Hiroshima in Iran” (International Security 42(1), 2017—no work of theirs titled “Reviving the Nuclear Taboo” exists in the record) reported a July 2015 experiment: against an Iranian scenario in which a ground invasion would cost 20,000 American lives, 55.6 percent preferred and 59.3 percent approved a nuclear strike killing 100,000 Iranian civilians; at 2,000,000 Iranian dead, preference stood at 47.7 percent and approval at about 59 percent; when diplomacy was offered as an option, about 41 percent chose it and about 40 percent still chose the strike. Dill, Sagan, and Valentino, in “Kettles of Hawks” (Security Studies, 2022), ran the experiments across the United States, Britain, France, and Israel: majorities or near-majorities in each country endorsed nuclear use when it was the most effective option, support fell as civilian deaths rose, categorical prohibitions did little, and Israelis were the most hawkish of the four publics and the British the least, with the French and Americans roughly equal; the desk could not pin the paper’s country-level percentages from its tables and prints the finding without them. Joshua Schwartz, in International Security in the spring of 2024, reported four experiments, three American and one Indian: approval was not significantly lower for an ally’s or partner’s use than for one’s own government, majorities or near-majorities of Americans approved American, French, and Israeli strikes, and about two-fifths approved even Pakistani or Russian strikes. His conclusion, filed as his: the findings are “inconsistent with the existence of a nuclear taboo or strong non-use norm” in public opinion.

The pushback is also on the record, and it is about the instruments. Ward Carpenter and Erica Montgomery, in International Security in 2020, found that when the instrument frames the choice in law and norm terms, majorities oppose the targeting of civilians; much depends, their work shows, on question structure. Lauren Koch and Matthew Wells, in “Still Taboo?” (Journal of Global Security Studies, 2021), are cited by Tannenwald herself as finding opinion consistent with the taboo; the desk did not retrieve their toplines and prints the citation at that grade. Sukin, Rodriguez, and Herzog, in Perspectives on Politics in 2026, surveyed 27,250 respondents across 24 countries observationally and found publics holding support for deterrence and support for abolition at the same time—65 percent of Americans, in the Herzog, Baron, and Gibbons work they cite, wanted the weapons eliminated under the prohibition treaty—with 45 percent of all respondents strongly disagreeing that nuclear use can be morally justified, and 38 percent of Americans. The sentence the desk draws from the whole section is narrow: the public’s aversion is strong in the abstract, conditional in the concrete, and no instrument in the file asks the question on which everything would turn in an actual crisis, which is what a particular public would accept from a particular president on a particular day.

Two neighboring measurements are filed separately because they are not use-approval. The first is the retrospective judgment of Hiroshima and Nagasaki, a different question with its own instruments: Gallup measured 85 percent approval against 10 percent disapproval in August 1945, 53 percent in 1990, and a band of 53 to 59 percent through 2005; Pew measured 56 percent calling the bombings justified in 2015, and, in a survey fielded June 2 to 8, 2025, before that month’s Iran strikes, 35 percent justified, 31 percent not justified, and 33 percent not sure, with 69 percent saying the development of nuclear weapons made the world less safe. Sagan and Valentino themselves caution that declining retrospective approval is a “misleading guide” to opinion in a contemporary scenario. The second neighbor is expectation. In the same YouGov survey of October 30 to November 2, 2025, that returned 7 percent for first use, 46 percent said an American nuclear war within ten years is likely, 52 percent said the chance is greater than ten years ago, and 65 percent were concerned about personally experiencing nuclear war—against 69 percent in March 2022, which is to say essentially flat, and the desk prints the pair to prevent a false surge claim. The Atlantic Council’s Global Foresight 2025, an expert sample of 357 fielded in late November and early December 2024, found 48 percent expecting nuclear weapons to be used in the coming decade, up from 37 percent in the 2024 edition. Expectation and approval sit an order of magnitude apart inside single instruments. A public can expect a thing it does not approve, and in the current measurements, it does.

Two wartime instruments complete the section, and one absence. In March 2022, after the invasion of Ukraine, the Chicago Council measured support for American nuclear retaliation at 72 percent if the United States itself were nuclear-attacked, 54 percent if it were attacked with chemical or biological weapons, 44 percent if a NATO ally were nuclear-attacked, and 31 percent if an ally were attacked chemically or biologically; in a scenario of Russia attacking Poland or Germany, 37 percent supported an American nuclear response in total—15 percent unconditionally and 23 percent only if Russia used nuclear weapons first—against 60 percent opposed, including 33 percent who said never. The AP-NORC survey of the same month is a concern instrument only and contains no question on American use; it is not cited in this report for use. The absence: no general-population American poll of 2025 or 2026 asking approval of an American nuclear first strike on Iran specifically was found in the record, although that is the use the autumn’s discourse predicted. The nearest item is YouGov’s 7 percent general first-use figure, and the desk does not present it as Iran-specific.

4. The Testing Question: A Directive, a Clarification, and No Test

The sequence begins with a post. On October 29, 2025, American time—October 30 in Korea, hours before his meeting with Xi Jinping at Busan—President Trump wrote on Truth Social: “The United States has more Nuclear Weapons than any other country. This was accomplished, including a complete update and renovation of existing weapons, during my First Term in office… Because of other countries’ testing programs, I have instructed the Department of War to start testing our Nuclear Weapons on an equal basis. That process will begin immediately.” The desk grades the post’s premises separately from its instruction. The claim of the largest arsenal is contradicted by the SIPRI and Federation of American Scientists totals filed in Section 9, which put Russia’s total inventory ahead of America’s. The claim of a complete update and renovation accomplished in the first term is overstated against the program record: none of the Sentinel missile, the Columbia-class submarine, or the B-21 bomber was fielded between 2017 and 2021. Russia’s October 2025 events—the Burevestnik and Poseidon announcements filed in Section 6—were tests of delivery systems and propulsion, not explosive warhead tests. And explosive testing in the American government is the legal province of the Department of Energy and its National Nuclear Security Administration, not of the department the post named. The instruction itself is verified as stated; reporting by CNN on November 14 and 15, 2025, attributed to anonymous officials, holds that the post was not coordinated in advance with the National Security Council or the Departments of Defense and Energy, and that Energy officials prepared to discourage explosive testing. That reporting is filed as reported.

The President then said the plainer thing on television, and his Energy Secretary unsaid it in the other direction. On 60 Minutes, recorded October 31 and broadcast November 2, 2025, Mr. Trump said of Russia and China, “They test way underground where people don’t know exactly what’s happening with the test,” and, asked whether the United States is “going to start detonating nuclear weapons for testing,” answered: “I’m saying that we’re going to test nuclear weapons like other countries do, yes.” The embedded claims are graded as claims: that Russia and China are testing explosively in secret is claimed by the President, denied by both governments, and not confirmed by the Comprehensive Nuclear-Test-Ban Treaty Organization; that Pakistan is currently testing is claimed by him with no corroborating record, Pakistan’s last announced tests being in May 1998; that North Korea is “testing constantly” is true of missiles, with no confirmed North Korean explosive test since September 2017. On November 2 and 3, Secretary of Energy Chris Wright told Fox News: “the tests we’re talking about right now are systems tests. These are not nuclear explosions. These are what we call noncritical explosions.” Secretary of Defense Pete Hegseth, on December 7, 2025, repeated the formula, to test “on an equal basis,” without specifying explosive or non-explosive. The Congressional Research Service, in December 2025, recorded the state of the question in one sentence: “it is unclear what change of policy, if any, is to take place.”

What was done, through November 2026, is a verified negative. No American explosive nuclear test has occurred. The last was the shot named Divider, in Nevada, on September 23, 1992. The United States maintains a readiness to resume testing in 24 to 36 months, on the Congressional Research Service’s account, which is to say an October 2025 directive could not have produced a prompt test on the government’s own timeline. What did occur is the standing program: subcritical experiments at the Nevada National Security Site, 34 of them since 1992 as of May 2024, and a subsurface, non-nuclear chemical high-explosive experiment at the site’s P-Tunnel on September 30 and October 1, 2026, announced by the Department of Energy as detection research. The responses abroad are on the record. President Putin, at his Security Council on November 5, 2025, in a Kremlin transcript the desk verified, instructed his foreign and defense ministries and services to gather information and submit coordinated proposals on possible first steps toward preparing tests; Russia would be, he said, “obliged to take reciprocal measures” if the United States or any test-ban signatory tested; Defense Minister Belousov said the Novaya Zemlya range could host tests at short notice; spokesman Peskov said no deadline was set. China’s foreign ministry said the United States should abide by its test-ban obligations and its moratorium. No Russian or Chinese explosive test followed. In Congress, bills to restrain testing were introduced—the RESTRAIN Act and the No Nuclear Testing Act among them—and none was enacted through November 2026 in the sources the desk searched. At the United Nations First Committee on October 31, 2025, the United States cast the sole “no” vote on the annual resolution supporting the test-ban treaty and the moratorium, the later General Assembly tally standing 168 to 1 with 3 abstentions. American opinion on the question, for the measurements file: the Economist and YouGov, October 31 to November 3, 2025, 1,656 respondents, found 48 percent opposed to resuming American nuclear testing and 33 percent in support. The testing taboo is the taboo’s younger sibling—a moratorium, not a treaty in force—and its state at this filing is the same shape as the larger one’s: the words moved, the behavior did not.

One allegation from the exchange belongs in this section at its own grade. On February 6, 2026, at the Conference on Disarmament in Geneva, Under Secretary of State Thomas DiNanno said the United States was “aware” that China had conducted explosive nuclear tests, including preparations for yields “in the hundreds of tons,” citing a specific alleged test of June 22, 2020. Assistant Secretary Christopher Yeaw, at the Hudson Institute on February 17, added the instrument: a seismic station in Kazakhstan had detected a magnitude 2.75 event at 0918 GMT that day, some 450 miles from Lop Nur, with yield estimated at about 10 tons, or possibly hundreds if the shot was decoupled from the rock. The test-ban organization’s executive secretary, Robert Floyd, answered on February 6 and again on February 17: the international monitoring system had detected two very small seismic events twelve seconds apart that day, far below the roughly 500-ton level of confident identification, and the system “did not detect any event consistent with the characteristics of a nuclear weapon test explosion at that time,” the cause being unassessable. China’s ambassador Shen Jian and foreign ministry spokesman Lin Jian called the allegation “false narratives” and “completely groundless.” The State Department’s own April 2025 compliance report had recorded no new moratorium-adherence issues for Russia or China, and a CSIS satellite study of February 13, 2026, was inconclusive. The desk’s filing is the three-part one the record supports: an American claim, a Chinese denial, and a treaty organization’s non-confirmation. It is not a verified test, and it is not filed as one.

5. Russia and Ukraine: The Signaling Record Since February 2022

The war’s nuclear record opens with a sentence that does not contain the word. In his address of February 24, 2022, President Putin warned that Russia’s response to interference “will be immediate and will lead you to such consequences that you have never experienced in your history.” On February 27, he ordered the Russian army’s deterrence forces onto “a special regime of combat duty.” The Strategic Deterrence Forces include conventional components, and no named alert level exists in Russian practice; spokesman Peskov, on February 28, named remarks by the British foreign secretary, Liz Truss, as a trigger. The only documented physical correlate of the order is the Russian defense ministry’s statement of February 28, 2022, that duty shifts at the Strategic Missile Forces, the fleets, and Long-Range Aviation command posts had begun duty “with reinforced personnel”—a command-and-control survivability step, on the assessment of the analyst Pavel Podvig. The Federation of American Scientists’ Hans Kristensen recorded no public evidence of bombers loading or of unusual submarine dispersal, and Admiral Charles Richard of Strategic Command recommended no change in American posture in March 2022. Director of Central Intelligence William Burns said on May 7, 2022, that the agency saw no “practical evidence…of Russian planning for the deployment or even potential use of tactical nuclear weapons.” The desk records two absences that govern everything after: no open-source documentation has appeared of Russian warheads moved from central storage or mated to delivery vehicles at any point in the war, and no Russian statement formally rescinding the February 2022 order has been found; a single former officer’s account, given to the BBC in November 2024, claims base-level alert stood down after two to three weeks, and it is filed as claimed only.

The year’s texts, kept whole, are these. Foreign Minister Lavrov said on March 2, 2022, that a third world war “would be nuclear and destructive.” Spokesman Peskov told CNN on March 22 that use was possible “if it is an existential threat,” and told PBS on March 28 that “no one is thinking about using…even about [the] idea of using a nuclear weapon”; both statements stand, in tension, on the record. Lavrov on April 25 called the risk “serious, real,” said it “must not be underestimated,” and in the same remarks said it was not to be “artificially inflated”; the desk prints the whole position. Ramzan Kadyrov, on October 1, 2022, called by Telegram for “low-yield nuclear weapons” after the loss of Lyman; Peskov answered on October 3 that doctrine, “not emotions,” governs. President Putin, at Valdai on October 27, 2022, said there was “no need” to strike Ukraine, his own characterization of his position. And on September 21, 2022, in the mobilization address, he said: “we will certainly make use of all weapon systems available to us. This is not a bluff.” The passage sits inside a paragraph about nuclear blackmail and does not name nuclear weapons as the systems meant; converting it into an explicit promise of nuclear use goes beyond the text, and the desk does not convert it.

The autumn of 2022 carries the war’s closest approach, and its evidence is retrospective and one-sided. In October 2022, Defense Minister Shoigu telephoned his Western counterparts alleging Ukraine was preparing a “dirty bomb”; no evidence was presented, France, Britain, and the United States rejected the allegation jointly, and International Atomic Energy Agency inspectors visited the named sites and found, in November 2022, “no undeclared nuclear activities or material.” The allegation is filed as alleged, inspected, and not substantiated; the “evidence” image Russia circulated was a 2010 photograph from Slovenia’s radioactive-waste agency. Of the underlying fear in Washington: Director Burns said on September 7, 2024, beside MI6’s Richard Moore, that “there was a moment in the fall of 2022 when I think there was a genuine risk of the potential use of tactical nuclear weapons”; Burns met his Russian counterpart Naryshkin at Ankara on November 14, 2022; Secretary of State Blinken, in January 2025, put the probability’s rise “from 5 to 15%,” his retrospective estimate, published with no instrument. American reporting of the period states the United States saw no indicators of Russia mobilizing its nuclear forces. No Russian document, order, or deployment evidencing a fall 2022 decision to use a nuclear weapon has been made public. The desk files the episode as the record leaves it: a genuine fear in one capital, documented after the fact by its holders, and no corresponding document from the other.

Dmitry Medvedev is the signaling record’s loudest voice and holds no launch authority, and both halves of that sentence govern his weight in this file. As deputy chairman of Russia’s Security Council, his published statements are verified as his statements: on September 27, 2022, his hypothesis that NATO would not intervene even after a Russian nuclear use; on January 19, 2023, that “the loss of a nuclear power in a conventional war can provoke the beginning of a nuclear war”; in May 2023, that “a nuclear apocalypse is not only possible, it is quite probable”; on February 7, 2024, “ballistic and cruise missiles with special warheads… This is the notorious Apocalypse”; on February 18, 2024, that restoring 1991 borders would mean global war against “Kyiv, Berlin, London, and Washington”; on July 31, 2025, an invocation of the “Dead Hand” perimeter system, which drew the American submarine statement filed in Section 10. A “Medvedev April 2026” apocalypse quotation circulating this year was not found in high-grade sources and is not printed. The desk files Medvedev as what the record shows: a former president publishing escalation rhetoric on a schedule, whose statements are evidence about the discourse of threat, and not orders, doctrine, or deployments.

Ukraine’s side of the nuclear record is a record of renunciation under pressure, kept in its own texts. President Zelensky, at Munich on February 19, 2022, initiated consultations under the Budapest Memorandum and said that if they failed Ukraine would have “every right to believe” the memorandum was not working; stronger paraphrases of that speech, in which he is made to threaten acquiring nuclear weapons, are not in his text. At Brussels on October 17, 2024, he recounted telling then-candidate Trump that “either Ukraine will be a nuclear power—and that will be our protection—or we should have some kind of alliance,” and concluded, “That is why we choose NATO, not nuclear weapons”; his office stated within days that Ukraine is not building and does not intend to build a nuclear weapon and remains committed to the Non-Proliferation Treaty. An anonymous official’s claim to Bild that October, that Kyiv could produce a bomb “within a few weeks,” is filed as alleged, denied by Ukraine’s foreign ministry, with no program evidence produced by any party. In February 2025 Zelensky called the 1994 bargain “stupid, illogical, and very irresponsible” and argued guarantees are owed—an argument about guarantees, not the announcement of a program—and the American envoy Kellogg publicly dismissed reacquisition. The memorandum itself, signed December 5, 1994, is a political memorandum and not a ratified treaty, and its operative word is “assurances,” not “guarantees”; the last warheads left Ukraine in June 1996, Ukraine consulted under it in 2014 and again in February and March 2022, Russia declined to participate in 2022, and the Agency has made no finding of Ukrainian diversion at any point in the war. Former envoy Richard Grenell’s statement of March 25, 2025, that the weapons “belonged to Russia,” is his claim; the documented middle position of the historians is that Ukraine held physical custody of about 1,900 strategic and 2,500 tactical warheads in 1991 and 1992 without operational launch control.

6. The Doctrine of November 2024, Belarus, and the New Missiles

On November 19, 2024, Russia replaced its written nuclear doctrine, and the text is verified. The Fundamentals of State Policy in the Field of Nuclear Deterrence, Decree No. 991, superseding the decree of June 2, 2020, was previewed by President Putin on September 25, 2024—“the use of nuclear forces is the last resort measure”—and published two days after reports of American authorization for ATACMS strikes into Russia and on the day of the first reported such strike; the sequence is documented, and the motive is the parties’ claims. The textual changes against 2020 are seven. The conventional-aggression trigger moved from aggression “when the very existence of the state is in jeopardy” to aggression that “creates a critical threat to their sovereignty and/or territorial integrity,” with sovereignty undefined and official usage, the Carnegie Endowment notes, extending toward regime stability—an analyst’s gloss, filed as one. Belarus and the Union State are named expressly. A new trigger covers the “massive launch (take-off) of aerospace attack weapons…and their crossing of the state border.” A non-nuclear state’s aggression “with the participation or support of a nuclear state is considered as their joint attack.” The deterrence formula lost the word “exclusively,” the danger list grew from six items to ten, and the arms-control reference was removed. A new section 20 states what was practice: “The decision to use nuclear weapons shall be made by the President of the Russian Federation.” Whether the operational threshold was lowered or existing guidance merely codified is disputed among the analysts—lowered, on the NIPP assessment; primarily signaling, on Carnegie’s—and the Defense Intelligence Agency’s 2025 assessment, carried by the Congressional Research Service, holds Russia “very unlikely” to use nuclear weapons in Ukraine unless its leadership judged an existential threat to the regime.

Belarus is the file’s cleanest case of the grading discipline. The announcement came on March 25, 2023, when President Putin told Rossiya-24 that, at President Lukashenko’s request—his account—Iskander-M missiles had been transferred, ten Belarusian aircraft re-equipped, training to begin April 3, storage to be complete by July 1, and control to remain Russian, framed as mirroring American nuclear sharing in NATO. The reversal is documented: a Russian diplomat had told the United Nations in October 2022 that there were “no plans,” and the Xi–Putin joint statement of March 21, 2023, four days before the announcement, said nuclear states should “refrain from deploying nuclear weapons abroad.” Mr. Putin said at St. Petersburg on June 16, 2023, that the “first” tactical weapons had arrived, giving no number; Mr. Lukashenko said on December 10, 2024, that the number was “dozens,” his figure, uncorroborated. What independent instruments verify is narrower and real: dual-capable delivery systems, an Iskander-M brigade at Asipovichy and Su-25 modifications and training, and storage construction consistent with a nuclear facility, a double-fenced depot near Asipovichy whose preparations the Federation of American Scientists judged in 2025 to be “nearing completion.” On warheads, the Federation’s Russian Nuclear Weapons 2025 and the SIPRI Yearbook converge, and the desk prints their sentence in substance: the leaders “repeatedly claimed” the deployment, but there is no conclusive visual evidence of warheads as of January 2026. Any specific number—“dozens,” or a figure near 130 that circulates—is unverified and is not used in this report. Custody, both governments state, remains Russian. The Oreshnik missile followed the same path into Belarus on statements of increasing solidity: Mr. Putin on August 1, 2025, said the first serial batch was in service and a Belarus deployment planned; Mr. Lukashenko on December 18, 2025, said the system had arrived the day before and was entering combat service; the Russian defense ministry released video on December 30, 2025, said to show Oreshnik on combat duty in Belarus; the Zapad-2025 exercise rehearsed planning for nuclear weapons and Oreshnik use, on the Belarusian defense minister’s account; and a September 7, 2026, claim by Ukraine’s intelligence chief of Oreshnik units in Belarus stands as a single-source Ukrainian claim, unconfirmed. The context number is an estimate and is printed as one: Russia’s non-strategic stockpile is put by SIPRI and the Federation at 1,000 to 2,000 warheads in central storage, a figure routinely miscited elsewhere as a deployed count.

The Oreshnik’s combat record is three uses, at three grades. The first, on November 21, 2024, against the Pivdenmash plant at Dnipro from Kapustin Yar, some 800 kilometers, is a verified event: the Pentagon’s Sabrina Singh described an experimental intermediate-range ballistic missile “based on Russia’s RS-26 Rubezh”; Ukraine claimed six warheads carrying six submunitions each at Mach 11, its claim, the submunition count not independently verified; satellite assessments found the damage limited. It was the first combat use of a multiple-warhead ballistic missile and of an intermediate-range ballistic missile in war. Russia notified the United States about thirty minutes before launch through the nuclear risk-reduction channels, on the statements of both governments—a fact the desk underlines, because the notification machinery worked in the middle of the escalation it exists for. Mr. Putin, televised that evening, called it a test of “one of Russia’s latest medium-range missile systems—in this case, carrying a non-nuclear hypersonic ballistic missile…named Oreshnik,” and said Russia considered itself “entitled to use our weapons against military facilities of those countries that allow to use their weapons against our facilities.” The NATO–Ukraine Council answered on November 26 that it was “another attempt by Russia to terrorise the civilian population in Ukraine and intimidate those who support Ukraine.” The second use, on the night of January 8 and 9, 2026, against the Lviv region, is verified as an event by three parties: British Defence Intelligence put the flight at 1,622 kilometers and the Ukrainian air force tracked about 13,000 kilometers an hour; the Russian defense ministry confirmed the strike, named the Lviv State Aircraft Repair Plant as the target, and gave as its rationale an alleged Ukrainian drone attack on Mr. Putin’s Novgorod residence of December 29, 2025—a premise Ukraine denied, American and European officials disputed, and British Defence Intelligence called “inaccurate”; the actual aimpoint is not established in open sources, and no casualties were reported. The third use, on May 24, 2026, stands at the grade of documented with weaker sourcing, treated in this report as probable: two missiles in a mass overnight attack, one striking private garages at Bila Tserkva and one falling in Russian-occupied Donetsk region, which Mr. Putin said on June 4, 2026, was intentional, against his own ministry’s prior account, and which President Zelensky called a failed launch. Production and inventory are claimed by Russia and estimated against it: Ukrainian intelligence in January 2026 said “no more than three or four,” and British Defence Intelligence said Russia “highly likely only has a handful.”

The Burevestnik and Poseidon announcements of October 2025 are Russian claims, graded as claims. General Gerasimov reported to Mr. Putin on October 26, 2025, a Burevestnik flight test of October 21 covering 14,000 kilometers in about fifteen hours; those figures are Russian official claims, published without telemetry. Partial corroboration exists and is narrower than the claim: Norwegian intelligence confirmed a test occurred from Novaya Zemlya and that the missile flew “significantly further than before,” and Norway’s radiation authority detected no radiation spike as of October 26. Mr. Putin himself said the employment mode and infrastructure remained to be determined—not fielded, in the word the analysts at CSIS also used in November 2025—against a program history, since its announcement in March 2018, of at least thirteen known tests with two partial successes before this one. The Poseidon item is of the same grade: Mr. Putin announced on October 29, 2025, a first launch from a submarine with its reactor started, a Russian claim, and said on November 4 that next-generation nuclear-powered cruise missiles were in development. These two announcements are the factual floor under the American testing directive of the same week: delivery-system and propulsion tests, by a state that has never conducted a post-Soviet explosive test, answered in Washington as if they were warhead tests. The desk prints the elements at their grades and does not print the metaphor of blurring, which is a characterization and not a documented fact.

7. Iran: Two Wars, One Stockpile the Inspectors Cannot See

The June 2025 war is the file’s operational center, and its facts are documented. Israel opened Operation Rising Lion on June 13, 2025—about 200 jets, more than 330 munitions, more than 100 targets the first night, on the Israeli military’s figures—with Prime Minister Netanyahu’s stated rationale including that “Iran could produce a nuclear weapon in a very short time.” The United States followed on the night of June 21 and 22 with Operation Midnight Hammer, on the Pentagon’s briefing by Secretary Hegseth and General Caine: seven B-2 bombers, fourteen GBU-57 bombs, twelve on Fordow and two on Natanz, Tomahawk missiles on Isfahan, seventy-five precision-guided weapons in total, and the first operational use of the GBU-57; the Tomahawk count conflicts in reporting, at about 24 against about 30, and is unresolved. President Trump said that evening that the facilities had been “completely and totally obliterated,” a claim made before assessment; General Caine’s materially narrower words were “extremely severe damage and destruction,” initial battle-damage assessment only, and he did not say obliterated. Iran struck the Al Udeid base in Qatar on June 23 with advance warning and no casualties, on which all sides agree, while the counts conflict—Iran’s fourteen launched and six hits, Qatar’s thirteen intercepted and one impact, other reporting’s nineteen launched, the bridge unknown. A ceasefire took hold June 24, 2025, and held until February 28, 2026.

The question the war was fought over was never reconciled, and the desk prints the conflict as the finding. Israel claimed a new, secret, accelerating weaponization program. The American intelligence community, in Director of National Intelligence Gabbard’s Senate testimony of March 25, 2025, assessed that “Iran is not building a nuclear weapon and Supreme Leader Khamenei has not authorized the nuclear weapons program that he suspended in 2003”—and noted, in the same testimony, “an erosion of a decades-long taboo in Iran on discussing nuclear weapons in public,” a sentence this report files as the other taboo in the Iranian file. The International Atomic Energy Agency’s Board finding of June 12, 2025, document GOV/2025/38, was a finding of safeguards non-compliance, the first in about twenty years, carried 19 to 3 with 11 abstentions; it was not a finding of a weapons program, and Director General Grossi told Sky News on June 18 that “we could not affirm that there is any systematic effort in Iran to manufacture a nuclear weapon.” No public release of evidence reconciles the three positions. The same conflict recurred, unreconciled, before February 2026.

The damage assessments form a graded sequence with no reconciliation available, and the sequence is the report. The Defense Intelligence Agency’s preliminary assessment, leaked June 24, 2025—preliminary and low-confidence by the agency’s own description, made about a day after the strikes—held that core components were not destroyed and the setback ran to “a few months,” with some briefed sources putting it at three to eight. Mr. Grossi, recorded June 27, said the damage was severe but not total and that a few cascades could spin again “in a matter of months,” a measure of resumption, not of rebuilding. The Pentagon, on July 2, said degraded by one to two years. Director of Central Intelligence Ratcliffe, on June 25, said “severely damaged…would have to be rebuilt over the course of years.” The Israel Atomic Energy Commission, the same day, said Fordow was “inoperable” and the setback “many years.” The Institute for Science and International Security, on September 8, 2025, working from Agency reports and satellite imagery, assessed all of Iran’s roughly 22,000 installed centrifuges destroyed or inoperative and wrote that “for the first time in over 15 years, Iran has no identifiable route to produce weapon-grade uranium in its centrifuge plants”—an institute’s assessment, not an Agency finding. A partial bridge stands in the record: the assessments measure different things, restart against rebuild against weapons-grade production. The physical bridge does not: no inspector has entered the underground halls at Fordow or Natanz, then or since. The DIA’s director, Lieutenant General Kruse, was removed on August 22, 2025; a linkage to the leaked assessment is alleged, not documented. Director Gabbard, at a hearing on March 18, 2026, stated that the intelligence community assessed the strikes had “obliterated” Iran’s enrichment program.

The load-bearing numbers of the Iranian file are a baseline, and the desk states their date every time they are used. The Agency’s report GOV/2025/50 of September 3, 2025, gives, as of June 13, 2025, its last estimate: total enriched uranium of 9,874.9 kilograms, of which 440.9 kilograms was enriched up to 60 percent U-235, 432.9 of it verified; the prior report had put the 60 percent stock at 408.6 kilograms as of May 17, 2025. Pre-strike movement is documented: eighteen blue containers entered a tunnel at Isfahan on June 9, 2025, in satellite imagery cited by the Agency and the Associated Press, and Iran’s foreign minister wrote the Agency on June 13 that Iran would “adopt special measures to protect our nuclear equipment and materials.” Every Agency report since—GOV/2025/65, GOV/2026/8 of February 27, 2026, GOV/2026/33 of June 4, 2026, and GOV/2026/50 of September 1, 2026—restates the June 13, 2025, estimate unchanged. The 440.9 kilograms is a June 2025 baseline, not a current measurement, and any present-tense use of it is an error this report does not make. The Agency’s position is not that the material is buried and halted; it is that the Agency cannot verify the stockpile’s size, location, or condition, or whether enrichment has stopped. On the material’s fate, the claims overlap and none verifies another: Mr. Grossi told the Associated Press on April 28 and 29, 2026, as a stated belief, expressly unverified, that the majority of the highly enriched uranium was at Isfahan, roughly 200 kilograms of the 60 percent material in its tunnels, seals unverified; the Institute, in June 2026, put “most of the HEU, but less than 70 percent,” at Isfahan; an American claim, attributed to Director Ratcliffe’s closed briefing as described to NBC, holds most of it “buried” at Isfahan and Fordow and extraction “extremely difficult”; President Trump said on Fox News on June 29, 2025, that he did not believe Iran had moved it; Iran gives no accounting, and its SNSC secretary Rezaei said on October 20, 2026, that “not a single gram of enriched uranium will be removed from Iran,” a negotiating position. The bridge among these accounts is unknown.

Access ended in June 2025 and has not resumed, and the Board’s escalation is dated. The Agency has had no access to any affected enrichment or conversion facility since June 13, 2025; its inspectors were withdrawn at the end of that June, and verification continues only at Bushehr. Iran’s suspension law, passed by the Majlis on June 25, 2025, and signed by President Pezeshkian on July 2, admits inspectors only with the approval of the Supreme National Security Council. A modalities agreement signed at Cairo on September 9, 2025, was never published and was described differently by its two principals; Iran formally terminated it on November 20, 2025, after the Board’s resolution of that day, GOV/2025/71, carried 19 to 3 with 12 abstentions. On September 9, 2026, the Board voted 23 to 3, with Russia, China, and Niger against and eight abstentions, to report Iran to the United Nations Security Council for safeguards non-compliance, the first such referral in twenty years, on Reuters’ account. GOV/2026/50 records “more than a year since the Agency lost continuity of knowledge” over the inventories. Mr. Grossi told the Security Council on October 27, 2026, that “since June 2025, we have not identified any activity to repair or reconstruct any of the facilities damaged in the military attacks of 2025 and 2026”—a satellite instrument, not an on-site one—and that the Agency still “has not been able to verify that there has been no diversion.” Of the Isfahan Fuel Enrichment Plant, declared by Iran on June 12, 2025, the Agency has never been given access and does not know whether it contains nuclear material. The Arms Control Association, in October 2026, on open-source observation and not as an Agency finding, reported that Iran “does not appear to be excavating its bombed nuclear material or restarting activity that would indicate weaponization.” The European three triggered the snapback of pre-agreement Security Council resolutions on August 28, 2025; the resolutions were reinstated in late September, the effective date contested in authoritative sources between September 27, on the Congressional Research Service’s account, and September 28, on the European and Arms Control Association accounts, both printed; the legality is disputed by Iran, Russia, and China against the European three, and the effect is documented.

The second war is verified as an event, and its nuclear facts are these. On February 28, 2026, the American Operation Epic Fury and the Israeli Operation Roaring Lion struck Iran’s nuclear facilities, missile infrastructure, bases, command centers, and leadership in coordination. Ayatollah Ali Khamenei was killed in the opening strikes, Iranian state media confirming it on March 1; his son Mojtaba Khamenei was named Supreme Leader on March 8. The strikes came two days after a Geneva round the mediators called the most substantive in years. Iran retaliated across Israel and the Gulf, in True Promise IV, and moved on Hormuz shipping, declaring the strait closed on March 4. The Agency confirmed damage to entrance buildings of the underground Natanz plant on March 3, with no radiological consequence expected, and records that after February 28 it “stopped conducting verification activities in Iran in accordance with the NPT safeguards agreement.” The casualty figures in coverage are attributed and conflict: the Iranian Red Crescent’s count of at least 787 killed in Iran; a strike on the Minab girls’ school killing about 100 to 165 children, the figures in conflict by source; six American soldiers killed early, and a later Pentagon figure, cited in October 2026 polling coverage, of 21 American lives in total. Security Council Resolution 2817 of 2026 condemned Iran’s attacks on Gulf states and civilians and its interference with shipping. A Pakistan-mediated two-week ceasefire was announced April 7 and took effect April 8. On June 17, 2026, the Islamabad Memorandum was signed, fourteen points, mediated by Pakistan and Qatar, Mr. Trump signing at Versailles and Mr. Pezeshkian at Tehran. Its Article 8 is the nuclear settlement that was agreed and then not executed: “The Islamic Republic of Iran reaffirms that it shall not procure or develop nuclear weapons”; the stockpiled enriched material was to be disposed of by a mutually agreed mechanism, “with the minimum methodology to be down blending on-site, under the supervision of the IAEA”; and enrichment was to be discussed in the final deal, for which Article 3 allowed sixty days. The memorandum carried no quantified limit on enrichment, centrifuges, or stockpile. Technical talks met at Bürgenstock on June 21 and 28. The arrangement broke down on July 8, 2026, after Iranian attacks on Hormuz shipping; the Revolutionary Guard navy declared the strait closed on July 12, and the United States reimposed a naval blockade of Iranian ports on July 14. No final deal was reached within the sixty days or since, and sporadic exchanges continued into this month, strikes on Qeshm Island reported on November 2 and 3, 2026, claimed by American officials through the media and not independently verified.

Iran’s treaty status, at this filing, is a verified negative of the first importance. Iran remains a party to the Non-Proliferation Treaty. No Article X withdrawal notice exists in the record through November 9, 2026—despite a triple-urgency withdrawal bill of August and September 2025 that was not passed and a member’s bill of March 28, 2026; despite President Pezeshkian’s explicit disavowal of withdrawal on September 26, 2025, repeated in September 2026; despite a call by some 130 members after the September 9, 2026, referral and an open letter from 32 conservatives on September 28; despite a National Security Commission member’s estimate that a withdrawal vote would draw “perhaps no more than 70 votes”; and despite Secretary Rezaei’s statement on state television on October 31, 2026, that withdrawal “totally depends on Washington’s behavior,” with no decision taken—the highest level at which withdrawal has been threatened on the record, and still not a decision or a notice. Ten withdrawal proposals stand registered, on a member’s count of October 31; none is on the agenda and none has been voted. The religious edict stands in the same suspended state. Its documented texts are the late leader’s: a message to the Tehran disarmament conference, dated April 2010 in most scholarly citations and 2009 in some Iranian ones, the conflict unresolved, holding that “we consider the use of such weapons as haraam”—the operative sentence is about use—and his statement of October 9, 2019, that “both making and stockpiling atomic bombs are wrongful because using them is forbidden based on religion (haram).” No single published fatwa text in formal jurisprudential form banning production was found, and the edict’s scope is disputed on the record. Since February 28, 2026, the edict’s issuer is dead, and his successor has neither reaffirmed nor revoked it in any source the desk found through November 9, 2026. The executive cites the late leader’s edict; the foreign ministry spokesman said on November 2, 2026, that “the opposing side has assassinated the issuer of the fatwa forbidding the production and use of nuclear weapons.” Whether the prohibition is operative doctrine in November 2026 is asserted by the executive and unconfirmed by the office that owns it. The acquisition rhetoric around it is documented as rhetoric: Kamal Kharrazi’s statements of May and November 2024, that if Iran’s existence were threatened “we will be forced to change our nuclear doctrine” and that “we have the capability to build weapons and have no issue in this regard”; Ali Larijani’s of March 31 and April 1, 2025, that a wrong move would “force Iran to move towards that”; and, since October 2026, open advocacy in the Majlis—a member’s call on October 18 for Iran to build and test a bomb, a second’s statement on October 31 that “it is time for the Iranian nation to withdraw from the NPT treaty and change the nuclear doctrine,” a third’s claim on November 3 that Iran “may have already achieved this capability,” claimed and unverified against the Agency’s reporting and his own government’s position—with Secretary Rezaei’s qualification of October 31 traveling with all of it: Iran is debating whether to keep the ban if hostilities resume, and no decision has been taken. At the negotiating table the file closes stalled, not collapsed, in Tehran’s telling: a seven-condition plan conveyed through Qatar in September, never published as one official document; Foreign Minister Araghchi’s statement of October 21 that the United States “agreed to the conditions but wants additional provisions”; a personal letter from Vice President Vance to Secretary Rezaei in late October which Mr. Rezaei had read and not answered, and which Mr. Araghchi called, on October 31, a “positive step” carefully examined, with no plans for talks; Vice President Vance’s question, put through Reuters on October 6—“If you don’t want a nuclear weapon, then why do you need 60 percent enriched fuel?”—and President Pezeshkian’s answer to the underlying question, given to Fox News on September 24, 2026: asked whether Iran wants nuclear weapons, “Not at all. Not at all.”

The desk closes the Iranian section with the absence that governs it. No American or Israeli nuclear-use rhetoric toward Iran was found in the record from June 2025 through November 2026. The nearest item is conventional: the deep tunnel complex at Pickaxe Mountain, Kuh-e Kolang Gaz La, a mile and a half to two kilometers south of Natanz, under construction since about 2020, never granted to Agency access, and not struck in either war. President Trump’s conventional threats against it are documented across July, October, and this month. On Newsmax on October 15 and 16, 2026, asked about the GBU-43/B, he said: “the mother of all bombs is not nuclear… No, it’s not nuclear at all. This is just an extremely large bomb… It’s like a tiny nuclear bomb [in effect]”; and of Pickaxe, “Now, we’re looking at Pickaxe Mountain. We may have to go there… We’ll knock it out fast.” He said on October 8, 2026, that the United States “will not be attacking Iran at any time prior to the midterm elections,” and as of November 9 no strike on Pickaxe Mountain has been reported to have occurred. The allegation that nuclear weapons are under consideration for it enters this report in Section 11, at the grade of alleged.

8. The Wider Field: Pyongyang, Beijing, Europe, and the Subcontinent

North Korea’s position is stated in its constitution and repeated in its leader’s words. Kim Jong Un told the Supreme People’s Assembly on September 21, 2025, in the text carried by KCNA and reported by Reuters and the Associated Press: “If the United States drops the absurd obsession with denuclearising us and accepts reality, and wants genuine peaceful coexistence, there is no reason for us not to sit down with the United States”; “We will never give up our nuclear weapons”; and, “Personally, I still have fond memories of U.S. President Trump.” The official English text adds that denuclearization is “the last, last thing to expect from us,” and possession stands enshrined in the constitution since the September 2023 amendment as “sacred and absolute.” The Nuclear Forces Policy Law of September 2022 provides for automatic and preemptive use in enumerated conditions, including attack on the leadership or strategic assets; the law’s primary text was not retrieved by the desk and is graded documented-as-reported. On September 26 and 27, 2025, meeting his nuclear scientists, Mr. Kim reviewed the expansion of fissile-material and warhead production and called for sharpening the “nuclear shield and sword”; the doctrine language includes a “second mission” beyond deterrence, actual use if attacked or judged under imminent attack. The estimates are attributed: SIPRI counted about 50 assembled warheads in January 2025 with fissile material for up to 40 more, and about 60 assembled in January 2026 with material for at least 30 more, a SIPRI summary page phrasing total potential as at least 90, both printed; the Federation is concordant; a KIDA presentation’s 127 to 150 and a South Korean defense minister’s 80 to 120 of August 2026 are documented as existing outliers and not adopted. President Trump’s statement of August 20, 2026, that North Korea had “57 very powerful nuclear weapons” is filed as claimed, inside the SIPRI and Federation range and with a precision those instruments do not carry. What was done: no North Korean explosive test in 2025 or 2026, and missile tests continued.

China’s numbers are other people’s estimates, and its doctrine is its own document. China publishes no stockpile figure. The Defense Department’s China Military Power Report of 2025, released December 23, 2025, puts operational warheads “in the low 600s through 2024,” holds China “on track to have over 1,000 warheads by 2030,” counts more than 100 silos at three fields likely loaded with DF-31-class missiles supporting an early-warning counterstrike posture, and counts 550 launchers and 400 missiles; the 1,500-by-2035 projection of the 2022 report was not repeated. SIPRI counts 600 warheads in January 2025 and 620 in January 2026, calls China’s arsenal the fastest-expanding of any country’s, and records hundreds of missiles loaded into the three northern silo fields by January 2026 with thirty silos being completed in the east; its thousand-by-2030 figure is a conditional projection, filed as claimed. Against the buildup stands the doctrine, declared since October 16, 1964: no first use. China’s white paper of November 27, 2025, China’s Arms Control, Disarmament, and Nonproliferation in the New Era—verified as a document, its first comprehensive arms-control white paper since 2005—reaffirmed no-first-use and a self-defensive strategy, reaffirmed support for the test-ban treaty’s “purposes and objectives” and for the moratorium, held its forces at the “minimum level required for national security,” and rejected the arms-race characterization. Western analysts, the Defense Department among them on the counterstrike posture and low-yield options, question the practical consistency of the two; that is an analytical assessment, not a Chinese doctrinal change, and no Chinese statement abandoning or qualifying no-first-use in 2025 or 2026 was found. On trilateral arms control, Beijing’s refusal is consistent and on the record: spokeswoman Mao Ning said on November 7, 2025, that the forces are “not on the same scale” and talks would be “unfair, unreasonable, and impractical”; the white paper assigns the United States and Russia “special and primary responsibility”; and at the 2026 review conference, Sun Xiaobo called the American proposal “in essence an attempt…to shift its special and primary responsibilities…onto others.” China urged the United States on February 3, 2026, to accept Russia’s one-year New START offer.

Europe’s movement is from debate to documents, and every document keeps the decision national. Chancellor-elect Friedrich Merz said on February 21, 2025, that “we need to have discussions with both the British and the French…about whether nuclear sharing, or at least nuclear security from the UK and France, could also apply to us,” and repeatedly ruled out a German national weapon, which the Non-Proliferation Treaty and the Two Plus Four Treaty bar. President Macron, in a televised address on March 5, 2025, opened “the strategic debate on the protection of our allies on the European continent through our (nuclear) deterrence,” with the decision, he said, one that “has always been and will remain in the hands of the President of the Republic.” Prime Minister Tusk told the Sejm on March 7, 2025, that Poland must “reach for the most modern capabilities also related to nuclear weapons” and was “talking seriously with the French”; no Polish weapon program followed. The Northwood Declaration of July 10, 2025, signed by Prime Minister Starmer and President Macron, states that the British and French forces “are independent, but can be coordinated” and that “there is no extreme threat to Europe that would not prompt a response by our two nations”; it establishes a Nuclear Steering Group, which first met at Paris on December 10, 2025, and it is not joint command, shared targeting, or pooled launch authority. Britain announced at the June 2025 NATO summit the purchase of at least twelve F-35A dual-capable aircraft and its joining of NATO’s nuclear mission, a procurement decision; the aircraft were not delivered in the 2025 and 2026 record. A Franco-German Nuclear Steering Group followed the Munich Security Conference of February 2026, Mr. Merz confirming on February 13 that confidential talks with Mr. Macron had begun. On March 2, 2026, at Île Longue, President Macron announced “forward deterrence”: structured dialogues with willing partners, allied participation in Strategic Air Forces exercises, possible forward deployment of nuclear-capable Rafale aircraft outside French territory, and allied conventional support—together with an increase in the number of French warheads, the first publicly announced increase since 1992, its number not disclosed; an end to public disclosure of the arsenal’s size; and no shared decision-making and no explicit guarantee, sole French presidential decision expressly retained. By June 2026, on the count of a CSS ETH Zurich policy brief of July 2026, nine states stood in structured bilateral discussions: Britain, Germany, Belgium, the Netherlands, Denmark, Poland, Sweden, Greece, and Norway. Remarks by Presidents Nawrocki and Merz in February 2026 about sharing are reported; no sharing agreement concluded through November 2026 was found. France’s arsenal before non-disclosure stood at 290 warheads on SIPRI’s January 2025 count. The desk files the divergence the texts require: the press shorthand of a European umbrella, sharing, or shield describes coordination among national forces under national decisions, and the documents themselves preserve sole national launch authority in every case.

The India–Pakistan crisis of May 2025 is the file’s proof that the signaling discipline can hold under fire. After the Pahalgam attack of April 22, 2025, which killed 26—India attributing it to Pakistan-based militants, Pakistan denying state involvement—India opened Operation Sindoor on May 7, and four days of reciprocal air, drone, and missile exchanges followed, the most extensive between the two states since 1971, ending in a ceasefire understanding on May 10. The nuclear signaling of the crisis is three items, at three grades. The Pakistan National Command Authority episode of May 10 was reported and then denied: state media reported a meeting called, and Defense Minister Asif told Geo News, “No meeting has happened…nor is any such meeting scheduled”; whether a meeting was called and cancelled or never called is not established. The Kirana Hills strike and radiation rumor is alleged and not verified: no Agency statement, no official confirmation, no radiological evidence. India framed Sindoor as conventional and calibrated and repeated, including at the General Assembly in September 2025, that it would not bow to “nuclear blackmail.” No nuclear weapon was used, threatened by name by either head of government, or moved in a publicly verified manner during the crisis. The mediation record is contested in the open: American diplomatic engagement on May 9 and 10 is documented on the State Department’s record of Secretary Rubio’s calls; India states categorically that the halt was a bilateral directors-general agreement with no mediation; Pakistan thanked President Trump. Mr. Trump’s figures—25 million, then 35 million, then 50 million lives; 250 percent tariffs; five aircraft, then eleven—are filed as claimed, uncorroborated, and changed across repetitions. SIPRI’s sentence on the crisis is the desk’s: “both sides took steps to avoid escalation.”

The prohibition treaty grows everywhere except where the weapons are. The Treaty on the Prohibition of Nuclear Weapons, adopted July 7, 2017, and in force since January 22, 2021, stands, on the United Nations Treaty Collection count of 2026, at 96 signatories and 75 states parties—70 ratifications and 5 accessions on the September 2026 instrument counts, Tonga’s accession of July 7, 2026, making it the 75th party and the 100th state to take formal action. Ghana’s ratification and Kyrgyzstan’s signature of September 26, 2025, brought the combined total to 99. No nuclear-armed state has joined, signed, or ratified; no NATO member has joined; and no change in any nuclear state’s position toward the treaty was found in 2025 or 2026. Its first Review Conference is scheduled for November 30 to December 4, 2026, at New York, in the future at this filing. Two older items from the rhetoric file close the section. In November 2023, Israel’s Heritage Minister Amichay Eliyahu—a junior minister with no nuclear authority, not a member of the war or security cabinet—said on Radio Kol Barama that dropping “some kind of atomic bomb” on Gaza was “one option”; the Prime Minister’s office called the remark “detached from reality” the same day, Mr. Eliyahu was suspended from cabinet meetings, and he later called the remark “metaphorical.” Israel has never officially confirmed possession, and SIPRI and the Federation estimate about 90 warheads. The remark is not Israeli policy and is not filed as such. Beyond President Putin’s standing conditional statements and doctrine, Mr. Kim’s second-mission language, and President Trump’s statements, no other head-of-government explicit nuclear-use threat in 2025 or 2026 meeting the primary-record standard was found.

9. The Architecture: What Lapsed, and What Still Binds

The treaty record of the decade is a list of endings, and New START’s is the load-bearing one. The treaty was signed at Prague on April 8, 2010, entered into force on February 5, 2011, and limited each side to 1,550 deployed strategic warheads, 700 deployed delivery vehicles, and 800 launchers and bombers; its single five-year extension was exercised in February 2021. On February 21, 2023, President Putin announced that Russia “suspends its participation,” expressly not withdrawing; the treaty contains no suspension clause, the United States holds the suspension legally invalid, Russia holds it a response to American, British, and French actions, and both characterizations stand attributed. The Russian foreign ministry said the same evening that Russia would continue observing the quantitative limits and continue the 1988 launch-notification exchanges. What stopped was the machinery: data exchanges, notifications, and telemetry ended, and inspections, paused for the pandemic in 2020, never resumed. American countermeasures of June 1, 2023, withheld American data and notifications and denied inspector visas and facilities while continuing to observe the central limits. The last agreed data exchange is September 2022’s: Russia at 1,549 warheads on 540 delivery vehicles. Both sides’ claims of continued observance since are unverifiable by the treaty regime, national technical means only. On February 5, 2026, the treaty expired by its terms; no further extension was legally available. The exchange around the expiry is on the record in full: Mr. Putin’s conditional offer of September 22, 2025, of one year of voluntary post-expiry observance of the central limits, without verification; press secretary Leavitt’s “sounds pretty good”; President Trump’s of October 5, 2025, “It sounds like a good idea to me,” with no formal acceptance ever delivered; his words to The New York Times of January 7 and 8, 2026, “If it expires, it expires. We’ll just do a better agreement”; his Truth Social post of February 5, 2026, rejecting extension, calling New START “badly negotiated…grossly violated”—a characterization that exceeds the published compliance record, which concerned inspections, data, and telemetry and not a finding that Russia exceeded 1,550, 700, and 800—and calling for a “new, improved, and modernized Treaty”; Foreign Minister Lavrov’s statement to the Duma on February 11, 2026, that Russia’s voluntary moratorium on exceeding the limits stands “only as long as the United States doesn’t exceed the said limits,” conditional, unverified, and restoring nothing; spokesman Peskov’s of February 6, that both sides “recognise the need to start talks…as soon as possible,” the subject raised in the Abu Dhabi Ukraine talks; the restoration of high-level American military-to-military dialogue on February 5 and 6; and the American position of February 6, from Under Secretary DiNanno and Secretary Rubio, that any successor must include China and cover all Russian systems. No successor talks are underway or planned, on the Arms Control Association’s account of April 2026, and an informal post-expiry no-upload understanding is not found: the Association wrote on February 20, 2026, that it “remains unclear,” and Assistant Secretary Yeaw said on February 17, “I know of no such agreement.” The upload potential is an estimate, the Federation’s: Russia could in theory raise deployed warheads by up to about 60 percent, and neither side had done so at scale as of SIPRI’s January 2026 count. The bottom line is documented and is printed as the section’s verdict because the instruments agree on it: for the first time since SALT I in 1972, no legally binding limit governs American and Russian strategic forces, and no bilateral inspection or data regime operates.

The rest of the architecture table is shorter because most of it is over. The INF Treaty died in 2019, American withdrawal effective August 2; Russia’s unilateral post-INF deployment moratorium was ended by its foreign ministry on August 4, 2025, Deputy Minister Ryabkov saying on August 10 that it was meant to “cool the heated heads” in NATO capitals, his claim of its effect. The Open Skies Treaty remains in force among its remaining parties, with the United States out since November 22, 2020, and Russia since December 18, 2021, and no American–Russian function. The CFE Treaty is functionally dead: Russia suspended in 2007 and withdrew effective November 7, 2023, and the United States and NATO suspended obligations toward Russia that November and December. The Vienna Document of 2011 is extant as a document and broken in operation between Russia and the West since 2022; a formal Russian suspension act and date in a primary OSCE source was not found, and the breakdown is graded documented with the legal act unverified. The test-ban treaty is not in force: the United States signed in 1996 and its Senate rejected ratification in 1999, a signatory only; Russia ratified in 2000 and de-ratified on November 2, 2023, a signatory only, stating it will maintain the moratorium unless the United States tests, and remaining in the treaty organization and its monitoring system; the counts of early 2026 are 187 signatures and 178 ratifications, with 35 of the required 44 Annex-2 ratifications. No American explosive test has occurred since September 23, 1992, no post-Soviet Russian explosive test has ever occurred, and China’s last was July 29, 1996, all verified. The Non-Proliferation Treaty is in force, with 191 parties on the Association’s count and 190 to 191 by instrument and date; its eleventh Review Conference, at New York from April 27 to May 22, 2026, under Ambassador Do Hung Viet of Vietnam, concluded without a final document—the third consecutive failure, after 2015 and 2022, no consensus document since 2010, the reported sticking points including language that Iran “can never seek, develop, or acquire” nuclear weapons—and the next conference falls in 2031. What still binds American and Russian nuclear forces, verified, is four things: the Non-Proliferation Treaty’s Articles I and VI, which carry no numerical limits; the Ballistic Missile Launch Notification Agreement of 1988, still operating, expressly confirmed by Russia on February 21, 2023, the channel through which the November 2024 Oreshnik notification ran and through which the Nuclear Risk Reduction Centers of 1987 still transmit; the Outer Space Treaty of 1967, barring nuclear weapons in orbit; and the two states’ test-ban signature obligations under Article 18 of the Vienna Convention, with their declared unilateral moratoria. The post-expiry restraint statements are unilateral, revocable, unverified political declarations, with no data exchange and no agreed counting rules, and are filed at exactly that weight.

The arsenals themselves, at January 2026, are counted by one team of analysts in two venues, and the desk says so before printing their numbers. The SIPRI Yearbook 2026, launched June 8, 2026, with its nuclear chapter by Hans Kristensen and Matt Korda, estimates a world total inventory of 12,187 warheads in January 2026, against 12,241 a year earlier, the decline coming solely from American and Russian dismantlement of retired warheads; military stockpiles of 9,745, up about 130; deployed warheads of 4,012, up from 3,912; and 2,100 to 2,200 warheads on high operational alert on ballistic missiles, nearly all Russian and American, with China and India now possibly deploying a small number of warheads mounted on missiles in peacetime on occasion. By country, total inventory against January 2025: Russia 5,420 against 5,459; the United States 5,042 against 5,177; China 620 against 600; France 370 against 290; Britain 225 against 225; India 190 against 180; Pakistan 170 against 170; North Korea about 60 against 50; Israel 90 against 90. Deployed, on SIPRI’s counting, which includes bomber weapons and is not a New START-accountable count: Russia 1,796, the United States 1,770. Stockpiles: Russia 4,400, the United States 3,700; the two together hold about 83 percent of stockpiled warheads and about 86 percent of the total inventory. SIPRI’s stated findings are its judgments and are filed as such: the states are “increasingly relying on nuclear weapons as instruments of national power—reversing decades of efforts to reduce the numbers and role of nuclear weapons—even as the risks of miscalculation and escalation are rising”; all nine nuclear-armed states modernized in 2025 and most deployed new systems; the inventory decline is driven only by dismantlement and “is likely to be reversed”; transparency is shrinking, France having said in March 2026 that it would stop disclosing its stockpile and Britain having stopped in 2021; and the significant increase in Russian non-strategic warheads predicted by the United States in 2020 “has still not materialized.” The Federation of American Scientists’ Status of World Nuclear Forces matches the Yearbook’s headline table—12,187 and 9,745; Russia 5,420 and 4,400; the United States 5,042 and 3,700; China 620—and the desk prints the warning that governs both: Kristensen and Korda author both publications, so the Federation and SIPRI in 2026 are substantially the same analyst team in two venues, not independent corroboration of each other. The Federation’s own Russian Nuclear Weapons 2025 puts Russia’s total stockpile at about 4,309 against SIPRI’s 4,400, both open-source estimates, printed together in the ledger below with the bridge unknown.

The Doomsday Clock is filed last in this section because it is a judgment and not a measurement, and the paper prints it as its makers’. On January 27, 2026, the Bulletin of the Atomic Scientists set its Clock at 85 seconds to midnight, the closest in its history—89 seconds in 2025, 90 in 2023 and 2024, 100 from 2020 to 2022, and 17 minutes in 1991, the farthest. The statement, edited by John Mecklin and presented at Washington by Alexandra Bell, gives as its ground a “failure of leadership”: three theatres of 2025 under the nuclear shadow, Russia’s war in Ukraine with its “lightly veiled Russian nuclear threats” and drone incursions into NATO states, the India–Pakistan crisis of May 2025, and the Israeli and American strikes on Iran’s facilities, of which the statement notes it remained unclear whether they constrained the program or persuaded Tehran toward covert pursuit, with the fate of more than 400 kilograms of 60 percent uranium unclear; a “full-blown arms race”; New START “set to expire, ending nearly 60 years of efforts to constrain nuclear competition”; proliferation pressure from doubts about American extended deterrence; the possible resumption of American explosive testing as a further accelerant; and, beyond the nuclear file, climate, biotechnology, artificial intelligence in defense, disinformation, and nationalistic autocracy as a “threat accelerant.” Eighty-five seconds is not a nuclear-only judgment, and the desk does not file it as one. It is the considered sentence of the institution that has kept the Clock since 1947, printed here because the taboo’s state is partly a matter of what the record’s keepers conclude from the record, and their conclusion is on it.

10. The American Posture: Modernization, Golden Dome, and the President’s Record

The standing American declaratory formula is a sentence from two prior administrations, and no new one has been published. The Nuclear Posture Reviews of 2018 and 2022 hold that the United States “would only consider the use of nuclear weapons in extreme circumstances to defend the vital interests of the United States or our allies and partners”; the 2022 review considered and rejected both no-first-use and sole-purpose declarations. No official restatement of the formula in 2025 or 2026 was found; the 2025 National Security Strategy and the January 2026 National Defense Strategy are, on the assessments of CSIS and SWP, “relatively light” on nuclear detail; and no new Nuclear Posture Review was published by the second Trump administration through November 2026. A classified Pentagon strategic review was reported in September 2026—Joseph Cirincione in the Bulletin, September 11—with no primary Defense Department release and no text, and it is not cited in this report as a published review. What the administration has said about the next architecture is on the record from February 6, 2026: Under Secretary DiNanno’s statement that the next arms-control era “will require the participation of more than just Russia,” with a “new nuclear architecture” announced including non-strategic weapons and China’s arsenal, on CSIS’s summary, and Secretary Rubio’s include-China argument of the same day.

The modernization program proceeds on numbers the government’s own auditors publish. The Congressional Budget Office, in April 2025, put the cost of the Defense and Energy Departments’ nuclear plans at $946 billion over 2025 to 2034, about $95 billion a year if carried out, a 25 percent increase over its 2023 figure of $756 billion. The Sentinel missile’s critical Nunn-McCurdy breach of January 18, 2024, took its acquisition cost up 81 percent to $140.9 billion; its Milestone B was rescinded, the program certified to continue on July 8, 2024, and the Government Accountability Office’s 2025 and 2026 accounts put the figure at no less than $141 billion, the first flight slipped about four years to March 2028, a new Milestone B targeted for the end of 2026 as a plan, and initial capability in the early 2030s. The Columbia class runs to twelve boats at $126.4 billion on the fiscal 2025 submission, its lead boat delayed twelve to seventeen months, its first patrol in late 2030 or 2031, and the Ohio class retiring at about one boat a year from 2027. The B-21 flew first on November 10, 2023, and a second test aircraft on September 11, 2025, with at least 100 planned; Air Force claims of July 2026 of a first operational aircraft at Ellsworth in 2027 are filed as claimed targets, no operational nuclear fielding being documented through November 2026. The low-yield W76-2 submarine warhead has been fielded since 2019. The sea-launched nuclear cruise missile, whose cancellation the 2022 review proposed, was directed by the fiscal 2024 defense law to proceed anyway, toward a capability not later than September 30, 2034—development, not fielding.

Golden Dome enters the nuclear file doctrinally, and its numbers diverge by construction. Executive Order 14186 of January 27, 2025, set the program in motion under the title “The Iron Dome for America.” The Congressional Research Service records the departure: the missile-defense reviews of 2019 and 2022 relied on deterrence against Russian and Chinese homeland strikes, and Golden Dome purports to add defense against those same delivery systems. The fiscal 2025 reconciliation law carries $24.4 billion for it. The cost figures diverge by scope and instrument and are never one number in this report: the President’s claim of May 2025, $175 billion over three years; General Guetlein’s later $185 billion; and estimates for a full space-based architecture running to about $1.2 trillion over twenty years, carried by SIPRI among others. No statement that Golden Dome’s interceptors are nuclear-armed was found. Russia and China’s joint statement of May 2025 holds that the program “completely negates” strategic stability, documented as their statement.

The President’s record, in both terms, is filed here as said against done, because the divergence is the data. In the first term: at Bedminster on August 8, 2017, North Korea would “be met with fire and fury like the world has never seen”; nothing was done in connection with the sentence. On January 2, 2018, answering Mr. Kim’s New Year address, he wrote: “I too have a Nuclear Button, but it is a much bigger & more powerful one than his, and my Button works!”—no literal button exists, a documented gloss the desk appends because the sentence is still quoted without it. In the 2016 campaign, at a Green Bay town hall on March 30, he declined to take nuclear use “off the table” for the Middle East or Europe; he did not affirmatively promise use, and the desk prints the refusal at its exact weight, while a second-hand adviser account of August 3, 2016, in which he is made to ask why the weapons cannot be used, stands as alleged, single-source, and unnamed. The 2018 review widened the threat framing, elaborated “significant non-nuclear strategic attacks,” and proposed the W76-2, fielded, and the sea-launched missile, not fielded in the term. The INF withdrawal was done in full: announced October 20, 2018, suspension and notice February 2, 2019, withdrawal effective August 2, 2019, the SSC-8 missile cited, Russia denying violation and suspending its own participation on July 3, 2019, and the first American test of a formerly prohibited ground-launched cruise missile on August 18, 2019. In May 2020, the Washington Post reported, deputies discussed a first explosive test since 1992 as trilateral leverage and the meeting ended without agreement; no test was done. The hurricane item of August 2019, in which the President was reported to have raised nuclear weapons against hurricanes, is filed as reported and alleged, denied by him on August 26, 2019—“I never said this. Just more FAKE NEWS!”—with the National Oceanic and Atmospheric Administration’s standing fact sheet documenting the science position.

The second term’s divergences are four, and each is documented at both ends. Testing: said, “begin immediately,” and detonations “like other countries do”; done, no test, no preparation milestone, the definition revised to non-explosive by his own Energy Secretary within four days, and his own readiness timeline standing at 24 to 36 months. Denuclearization: at Davos on January 23, 2025, “We’d like to see denuclearization… President Putin really liked the idea of cutting way back on nuclear… China would have come along too”—the characterizations of Mr. Putin and of China filed as claimed, spokesman Peskov saying on January 24 only that Russia wanted disarmament talks resumed “as soon as possible,” and no Chinese endorsement found; done, no trilateral negotiation convened through November 2026, and New START allowed to expire. Budgets: on February 13, 2025, in two venues, “let’s cut our military budget in half,” and “There’s no reason for us to be building brand new nuclear weapons. We already have so many you could destroy the world 50 times over, 100 times over,” and “President Putin and I agreed we were gonna do it,” the agreement claimed; done, no meeting, no budget halved, no program cancelled, and the $946 billion nuclear decade and $24.4 billion of Golden Dome funding proceeding. Submarines: on August 1, 2025, answering Mr. Medvedev, “I have ordered two Nuclear Submarines to be positioned in the appropriate regions”; the physical movement is claimed by the President and unverifiable by design, the Navy offering no comment. Two later statements close the record at their grades: the Newsmax claim of October 15, 2026—“We’re actually talking to Russia about that, and we’re talking to China about that denuclearization”—a presidential claim, uncorroborated; and a circulating story of a Shenzhen summit among Presidents Trump, Xi, and Putin, which was not found in high-grade sources and is not used.

11. The Argument in Public: Penn, Carlson, Jones, and the Answers on the Record

A special report files its occasion as record, and this report’s occasion is that the autumn of 2026 argued about nuclear use in public, at mass reach, for weeks. The desk files the argument the way it files everything else: what was said, by whom, when, in what words, and at what grade. No prediction in this section is argued true or false anywhere in this report. Their existence, wording, reach, and timing are the evidence, because a prohibition’s condition is partly measured by what can be said about breaking it, in whose voice, to how many, and with what answer.

The Sean Penn file is closed, and its value to this report is corrective. No item the desk retrieved has Mr. Penn using the phrase “nuclear taboo,” and no item has him, in his own quoted words, calling for a nuclear strike. The most direct item is the oldest: on Fox News’s Hannity on April 5, 2022, in a transcript corroborated by Variety and TheWrap, he said: “Even countries that have nuclear weapons can remain intimidated to use them, and we are seeing that now with our own country. And I fear what that legacy is going to be. Nobody wants to see a nuclear conflict.” And: “At the same time, if only one bully will be able to use those weapons as a threat, we have to rethink what we’re doing.” Partisan headlines of the period—that he said America must not be intimidated from using nuclear weapons against Russia, that he suggested nuclear weapons should be used—are filed as alleged glosses: the quoted words address intimidation and the monopoly of threat, and expressly disavow wanting a nuclear conflict. In a Newsweek interview published February 20 and 21, 2026, he said: “A lot of people in the media were initially afraid of a nuclear war. You got to be concerned about nuclear war every day, but what country are we when we are just going to be a punk to something we not only were warned about, but we…invented it? Like, ‘Oh, we’ll drop it on the Japanese, but as soon as it might come our way, we will hide and let whoever has to die, die.’” It is the only Penn item retrieved that invokes the American invention of the weapon and its use on Japan. On CBS’s Face the Nation on September 17, 2023, he said American caution was “a lean on in the fear of nuclear conflict…of course, is possible. And that’s to be concerning. The likelihood is extremely low. And as one of our witnesses in the film says, you know, are we going to let a gangster with nuclear weapons dictate the way we live?” and that “at some point, caution becomes cowardice.” On Club Random on June 15, 2025, on Mr. Putin’s threats: “I don’t know why our president hasn’t pre-announced a red line. And I don’t mean to talk a line. I mean a red line on, you know, don’t even talk about nuclear weapons.” The clips that circulated in August 2026 as if new are from Shane Smith Has Questions on VICE, released July 16, 2026, in which his nuclear wording, documented by clip transcription and corroborated by three clip summaries, is comparative: “A lot of people were critical of me early on about Ukraine, saying, ‘If we escalate, Russians are going to use a nuclear weapon.’ That’s possible any day with just loose nukes, forget about a state actor doing it, or if Pakistan sells something to somebody.” The word “bluff,” attached to him in the circulation, is not his word in the retrieved text; it is commenters’ gloss. A CNN appearance with Wolf Blitzer that circulates dated March 2026 is in fact March 1, 2023; no March 2026 appearance exists in the record retrieved, and his verified words that day are conventional: “They need more ammunition, more long-range precision weapons, and yes, fighter jets.” His 2023 film Superpower carries the Budapest bargain as a broken security obligation, in his Berlin and podcast remarks of that year—“a gangster with nuclear weapons”; a proposal in his own words that Ukraine acquire or reacquire nuclear weapons was not found, nor any statement by him responding to Mr. Putin’s February and March 2022 alerts, nor any 2022 to 2026 statement opposing nuclear weapons or advocating disarmament. The desk files the file at its sum: a prominent advocate arguing, across four years, that fear of nuclear use should not govern Western policy, in words that stop short of advocating use, inside a circulation that repeatedly carried him further than he went.

The Carlson allegation is the autumn’s central document, and it is alleged. On November 1 and 2, 2026, Tucker Carlson published on X that “under pressure from Israeli prime minister Benjamin Netanyahu, the Trump administration may be within days of launching a nuclear strike on Iran. The target would be Pickax Mountain in central Iran… No conventional weapon in the American arsenal can penetrate 300 feet of solid granite… The only military option is a nuclear strike… ‘Bibi calls Trump about it every day,’ says an informed observer”; that “after the first nuclear strike, World War Three is inevitable. That war would be short and final”; and that the decision would come “by November 7.” The post further alleges that Russians told two governments an American strike would free them to use nuclear weapons against Ukraine, and that the administration had “since February…quietly rewritten America’s nuclear doctrine”; no verifiable documentation accompanies either allegation in the post. The Daily Caller News Foundation states that it “independently corroborated the central allegation with a source familiar with the matter”; that corroboration claim is documented as the Foundation’s statement, and the allegation itself remains attributed to unnamed sources and is filed as alleged. Two timing facts stand without argument: every item in this run after October 15, 2026, is a prediction made against a standing on-record presidential denial, Mr. Trump having told Newsmax that day, “I wouldn’t use nuclear. We don’t have to”; and the November 7 date passed within this file’s window with no strike documented in any source the desk retrieved. On November 3, 2026, a Hannity interview renewed the President’s conventional threat against Pickaxe without nuclear wording in the recovered summary. Secretary Rubio has said no final decision on resuming strikes has been taken. Reporting by Axios of October 31 to November 2 holds that Mr. Netanyahu is pressing for conventional strikes after the November 3 midterms and the President prefers diplomacy, on anonymous sourcing, filed as reported.

The Jones run is documented at a density the desk has not previously filed for a single broadcaster. Alex Jones’s own site stated on November 2, 2026, that he “has warned for several months that POTUS has been considering launching nukes on Iran”; the dated instances support a continuous run from at least October 26 through November 8, 2026, and episodic instances from July 2026. Venue discipline first: Infowars in its prior form ended on May 1, 2026, with what its host billed as his “final InfoWars broadcast”; by the autumn the program is billed as The Alex Jones Show and Alex Jones Live on AlexJonesLive.com, Rumble, and his network accounts, and a November 2026 item filed simply as Infowars misnames the venue. The instances: October 26, a Sunday interview billed “Warfighter Warns Of Coming Crisis & Renewed War Going NUCLEAR,” with his own post that “Netanyahu Is Putting Into The Minds Of Israelis & Americans That Iran Is Going To Strike With A Nuclear Weapon, Therefore, Israel Must Strike First!”; October 28, a clip asserting that activation of the Pentagon’s reported “doomsday plan” could lead to “full-on nuclear war with Russia,” and his posts of October 28 and 29 that “The Government Is Warning Americans To Stock Up On 90 Days Of Food & Water For Potential Nuclear War” and that “DOOMSDAY Planes Are In The Air In A Massive Simulation For Nuclear War”; October 30, the site headline “MAJOR EVENT INCOMING: AI CYBER ATTACK, GLOBAL NUCLEAR WAR, OR BOTH!”; November 2, the convergence day, his posts that “Trump is preparing to launch a NUCLEAR sneak attack against Iran!” and that “my research and sources confirm that Tucker Carlson is 100% correct that President Trump is seriously considering the use of nuclear weapons against Iran’s enrichment sites,” with the further quoted matter of a “radioactive cloud hundreds of times more deadly than Chernobyl” and that “if Trump gives a green light for launching a offensive nuclear strike on ANY country, his cabinet must immediately remove him via the 25th amendment”; November 3, in a podcast-feed transcript, the nearest fully worded instance of the word this report’s title for the discourse avoids assuming: “they’ve leaked that they’ve been looking at tactical nuclear weapons to try to quote, take out their deep underground nuclear facilities like pickaxe mountain. So people trying to deny that and say that Tucker’s warning is bull want us to sleepwalk into Armageddon.” The desk dates the sentence November 3, and does not move it: for November 8 specifically, the word Armageddon in Mr. Jones’s mouth was not found. November 4 carried a guest segment billed with James Fishback’s words, that if the President publicly threatened a tactical nuclear weapon on Iran “He Should Be Removed From Office Via The 25th Amendment”; summaries conflict on whether the host endorsed or resisted the call in the segment, and his in-segment position is filed as claimed and conflicting. November 6 carried the headline “Iran Tells Population To Prepare For US Nuclear Strikes, Hands Out Iodine!” and, in the episode summaries, the probability sentence—“I think it’s about a 60% chance right now. I think it’s about a 10% chance he’s going to use nuclear weapons”—attributed by PodcastRex to Mr. Jones as his stated estimate, and the “genie” sentence to his guest, General Flynn: the President must say he will not consider nuclear use “because once that genie’s out of the bottle, it’ll never get back in.” November 8’s Sunday program existed under its own billing, verified on the program’s page: “More White House Sources Confirm President Trump Strongly Considering Imminent Nuclear Attack on Iran, TUNE IN NOW as Alex Jones Breaks Down the Latest Nightmare Developments, PLUS, Israeli Ambassador Confirms Takeover of US Military”—with the complication, on the same primary page, that “Royce White hosts this edition of Sunday Night Live!,” so the page alone cannot establish that Mr. Jones hosted every segment. A podcast-feed transcript fragment, from a third-party aggregator and not a primary archive, carries his voice at the opening: “Trump is strongly considering a first strike sneak attack on Iran. More White House sources have confirmed.” No complete primary recording or transcript of the November 8 program was recovered, and no Media Matters monitoring item for that date was found. The baseline against which the run is measured is two dated older instances, the cleaner being July 13, 2022: “If we don’t have a nuclear war in the next six months, we can count our lucky stars,” in a segment citing Minister Lavrov’s April 2022 warning; a third clean pre-2025 primary transcript was not found to that standard. The difference the record supports is one of form, not of truth: the 2022 instance is generic, a six-month window on a Ukraine trigger, and the October and November 2026 run is specific and near-daily—a named target, a decision date inherited from Mr. Carlson, a named constitutional mechanism urged, and repetition on October 26, 28, 29, and 30 and November 2, 3, 4, 6, and 8.

The comparable rhetoric and the counter-discourse complete the section, each at its grade. Representative Marjorie Taylor Greene published on X on August 16 and 17, 2026: “They are discussing using nuclear weapons on Iran in strategy meetings. Yes you read that correctly. It’s real. I’m not speculating, I know. And it’s pure evil,” and afterward that the President “may actually be the one who delivers a nuclear holocaust”; the underlying meeting claim is alleged, no evidence named, and the White House answered through Davis Ingle to USA Today that she “has no idea what she is talking about…should quit spreading Fake News.” Mrs. Greene and Candace Owens had both called for the 25th Amendment on April 7, 2026, answering the President’s post of that day—“A whole civilization will die tonight”—to which Mr. Jones answered on X that “the definition of genocide is destroying an entire civilization/people!” A Mark Levin sentence, “like Harry Truman, use the bombs,” on Iran, circulates as Mr. Jones’s attribution; no primary Levin transcript was recovered, and it is not found as primary. The Pakistani nuclear-retaliation claim repeated through the November items traces to June 2025 and to an Iranian official’s statement on Iranian state television that Islamabad had promised nuclear retaliation if Israel used nuclear weapons on Iran; Pakistan’s foreign minister dismissed it at the time, and it is filed as alleged at origin and denied by the alleged promiser, its repetitions being repetitions of a denied claim. Russian state television’s discourse runs in the other direction and is documented through its monitors: Vladimir Solovyov’s of February 1 and 2, 2026, that one nuclear detonation in space “solves” the Starlink problem; of August 17, 2026, “Hit them with Poseidon. Give the British three days to learn to breathe underwater”; of October 28, 2026, “this time the boot will be nuclear…what is there to do among the radioactive dust of Paris, Brussels, Berlin?”; and of November 4, 2026, “the need to create a gigantic security zone, including the use of nuclear weapons”; and Dmitry Kiselyov’s lead segment of March 15, 2026, titled “Nuclear Strike,” predicting a possible American or Israeli nuclear use against Iran and using a Carlson clip as its documentation—the direction noted, a Russian state host predicting American use. A clean dated 2025 or 2026 primary instance from Margarita Simonyan was not found; her documented instances are earlier, and a December 2025 circulation of Mr. Kiselyov is his 2019 segment recirculated, not a 2025 instance. The counter-discourse is named and dated: Robert Peters of Heritage, to the Daily Caller News Foundation on November 3, 2026, that “there is a near-zero chance we will use nuclear weapons on Iran”; Jeff Knopf of Middlebury, to the same outlet, that an American strike “would break a more-than-eighty-year tradition of not using nuclear weapons that has come to be seen as a taboo against nuclear use”; Philipp Bleek of the James Martin Center, that breaking the norm “would be a world-changing event…it might also have the opposite effect, galvanizing the world to bolster the norm”; Ian Bremmer’s judgment, carried in commentary of November 3, that Mr. Carlson’s post is “ludicrous and false”; the London Review’s “The Mountain Problem” of November 2, 2026, that “Carlson offered no independently verifiable evidence, and neither Washington nor Jerusalem has confirmed that a nuclear strike is under consideration,” with CSIS’s October analysis judging Pickaxe “probably not yet capable of enrichment,” the technical premise contested in the open literature; the Spearpoint Dispatch’s “STOP SELLING ARMAGEDDON!” of November 2 and 3, answering Mr. Jones and Mr. Carlson by name; Liz Wheeler’s of November 8, 2026, “TUCKER CARLSON LIES ABOUT TRUMP”; and the European Leadership Network’s Protecting the NPT of May 2026: “the nuclear taboo has held for eighty years… But this very success has bred complacency.” Reach, in this section, is measured only with its instrument named: Mr. Jones’s X account displayed 4.4 million followers in November 2026 platform counts, against 2.3 million displayed in the bankruptcy reporting of June and July 2024, a displayed-count change and not an audited audience; his network’s Facebook page displays 190,000 followers; web-traffic estimates conflict by instrument, SimilarWeb’s March 2026 snapshot giving about five million visits that month for his former domain against mid-2026 mirror estimates near 560,000 monthly, both printed with the bridge unknown; his in-program claim of November 3 to “over 400 radio stations” is filed as claimed, no affiliate list recovered; and per-episode audience figures for any October or November 2026 episode were not found from any named instrument.

12. The Contradictions Ledger: Where the Record Does Not Agree with Itself

The desk closes the file the way the record leaves it: with its disagreements printed, not resolved. First, the public: majorities that say only-in-response or never, in the abstract instruments across twenty-one years, and majorities that approve concrete strikes, in the scenario experiments across thirteen; both measured, neither reconciled, and which one is the taboo is the literature’s open question, not this report’s to settle. Second, the arsenals: the Federation’s Russian Nuclear Weapons 2025 puts Russia’s stockpile at about 4,309 warheads and SIPRI’s Yearbook at 4,400, both open-source estimates from an overlapping team, printed together; the Federation’s running text says about 3,912 deployed against its own table’s about 4,012 including about 100 non-strategic weapons, and its historical paragraph says 12,321 warheads at the beginning of 2026 against the same page’s headline of 12,187, a superseded vintage on the page’s own warning. India’s deployed warheads appear as about twelve in SIPRI-derived reporting for January 2026, the first such figure, and as zero in the Federation’s table with an SSBN-carriage footnote; the bridge is unknown and is likely a classification difference, the likeliness being the desk’s inference, flagged. Third, the Iran damage: “a few months” against “many years,” from agencies of the same two governments within the same week, the assessments measuring different quantities and the underground halls unentered by any inspector, then or since. Fourth, the counts inside single events: fourteen GBU-57s agreed and Tomahawks at 24 against 30; fourteen missiles at Al Udeid with six hits, thirteen intercepted with one impact, nineteen launched. Fifth, the snapback’s effective date: September 27 on the Congressional Research Service’s account, September 28 on the European governments’ and the Arms Control Association’s. Sixth, the small numbers that carry larger filings: the Bulletin’s own statement gives atmospheric carbon dioxide at 150 percent of preindustrial in its summary and 152 percent of 1750 levels in its climate section; the Human Security Lab’s sole-authority figure stands at 32 percent in the university’s release and 30 percent in the Bulletin’s write-up, on different wordings; the sole-authority survey itself is a special population, 750 veterans and active-duty respondents, and is not filed in this report as a general-population measurement. Seventh, General MacArthur’s requested bombs: 26, 34, and 38, by source, from December 1950. Eighth, the Khamenei edict’s founding message: April 2010 in most scholarly citations, 2009 in some Iranian sources. Ninth and last, the report’s own method, stated once: where this file’s instruments conflict, the desk has printed both readings and named the bridge unknown rather than building one, because a bridge the record does not contain is a claim, and claims in this paper are labeled. The final paragraph of the report is the first paragraph’s restatement, as the form requires: no nuclear weapon has been used in war since August 9, 1945; the prohibition that record embodies is measured as strong in principle and conditional in the concrete, defended in print by scholars who call it fragile and by none who call it strengthening; it survived a testing directive answered by no test, a doctrine revision answered by no use, two wars against a threshold state that neither crossed the line nor left the treaty, and the lapse of the last strategic arms treaty into a silence of unilateral, unverified restraint; and through the autumn of 2026 it was argued about, at mass reach and in apocalyptic vocabulary, by broadcasters whose central allegation stands unverified against a presidential denial on the record—the argument itself now part of the record the next filing will inherit.