nuclear.academy
Module 1

Nuclear Security: Why It Matters

What nuclear security is, where it came from, what actually stands between the world and more nuclear weapons, and why the coming decades will need more of it — drawn from a lecture given to first-year nuclear engineering students.

Learning objectives

This module is adapted from a lecture to first-year nuclear engineering students at the University of Tennessee. It keeps the lecture’s argument and most of its examples, drops the material that dates in months (which lives in the site’s current-situation pages), and adds the sources a reader would want to check. Each section ends with three questions you can use to explore the material yourself.

1. Why nuclear security, and why now

The world is building nuclear reactors again. After three decades in which the United States opened almost no new plants and much of the industrial base that had built the earlier ones quietly dispersed, electricity demand is rising, governments have relearned what carbon-free baseload power is worth, and dozens of reactor designs — large and small, familiar and novel — are moving from drawing board into licensing, and the first demonstration units have reached criticality. The people who built and ran the first generation are, meanwhile, leaving the workforce, and whether they are being replaced fast enough is an open question that the expansion makes urgent.

Every reactor, every kilogram of fuel, and every design idea in that expansion has a security dimension. That is the first thing to understand about the subject: nuclear security is not a specialty bolted onto the side of the nuclear enterprise by people in uniforms. It is built into the choice of fuel, the layout of a plant, the way material is weighed and tracked, the way data is handled, and the way people are vetted and trained. An engineer who does not think about it is making security decisions anyway — badly.

This module organizes the field around four questions. They are the author’s framework rather than an official taxonomy, but every job in the field answers at least one of them:

  1. How do we assure that nuclear material and technology are where they should be, doing what they should do, and properly protected?
  2. How do we detect activity outside the bounds of appropriate use?
  3. How do we respond effectively if something bad happens?
  4. How do we assess honestly what we know — and what we only think we know?

The first is the work of material control, accounting, and physical protection. The second is the work of safeguards inspectors, detector designers, and analysts. The third is the work of emergency response and nuclear forensics. The fourth is the hardest and the least glamorous, and it is the reason the other three ever improve.

Explore. (a) Which of the four questions does an international safeguards inspector mainly answer? (b) Why is a reactor designer “making security decisions anyway”? (c) The module says the expansion of nuclear power makes the workforce question urgent. What evidence would you want before accepting that, and where would you look for it?

2. Security, safety, safeguards, surety — and why the words get confused

Four words, four different jobs. The distinctions matter because the institutions, the law, and the engineering are different for each, and because in several of the world’s languages two of them share a single word.

Security is “the prevention and detection of, and response to, criminal or intentional unauthorized acts involving or directed at nuclear material, other radioactive material, associated facilities or associated activities” [1]. The threat is a person who intends harm. Protection of sensitive information and technology is part of the same job in practice, though it is handled under a different body of law.

Safety is “the achievement of proper operating conditions, prevention of accidents and mitigation of accident consequences, resulting in protection of workers, the public and the environment from undue radiation risks” [2]. The threat is a failure — of equipment, of procedure, of judgment — with nobody intending anything.

Safeguards is verification: the system of accounting, inspection, and analysis by which the International Atomic Energy Agency and national authorities confirm that nuclear material is being used peacefully and that a state is complying with its agreements. The concern is a state’s intent, and the tool is measurement [3].

Surety is a term used in the United States, by the Department of Defense and the Department of Energy, for the combined safety, security, and use-control measures that ensure a nuclear weapon will neither be lost, detonated without authorization, nor degraded [4]. It sits outside the civil enterprise but shares its vocabulary and much of its science.

Now the complication. English happens to have two words, safety and security, for two of these ideas. Of the six official languages of the United Nations — Chinese, English, French, Russian, and Spanish under General Assembly resolution 2(I) of 1946, with Arabic added by resolution 3190 (XXVIII) in 1973 [5] — French, like English, has two words, sûreté and sécurité. Spanish and Russian have one root word for both and add a qualifier to tell them apart: the IAEA’s Spanish glossary is the glossary of seguridad tecnológica (safety) and seguridad física (security), and its Russian usage is ядерная безопасность for safety and физическая ядерная безопасность for security. Chinese official usage now distinguishes 核安全 (safety) from 核安保 (security), although 安全 remains the everyday word for both. Arabic distinguishes amn (security) from aman (safety) [2]. A treaty, a training course, or a plant procedure written in one language and read in another can lose the distinction, and the confusion is not academic: a “safety culture” and a “security culture” ask different things of the same people, and a program that has one is often assumed, wrongly, to have the other.

A useful test for any measure you encounter: who is on the other side of it? If the answer is “nobody — physics, weather, or fatigue,” it is safety. If the answer is “someone who wants this to go wrong,” it is security. If the answer is “a government that has promised not to,” it is safeguards. The test is a simplification: a containment building or a guarded gate serves more than one purpose at once, and the point is to know which purpose a given requirement is written for.

Explore. (a) A reactor’s containment building protects against both an accident and an attack; which of the four words applies to each purpose? (b) Spanish and Russian use one root word for safety and security and add a qualifier. What could go wrong when a document written in one of those languages is read by someone who does not know the convention? (c) Which of the four concerns a state’s intentions rather than a person’s or a machine’s?

3. A short history: secrecy before the bomb, and what the Cold War taught

Security in the nuclear field is as old as fission. In the months after the discovery was published in early 1939, physicists in several countries recognized from the open literature what a chain reaction could mean, and some of them began, on their own initiative, to withhold results and to speak in code. The habit carried into the wartime program. Glenn Seaborg later recalled that the plutonium project — the element was discovered in 1941 and the Metallurgical Laboratory organized in 1942 — called the new element “copper,” which worked until the chemists needed real copper, at which point the metal became “honest-to-god copper” [6]. The point of the anecdote is not the joke but the instinct behind it: the people closest to the science understood at once that information about it was a strategic quantity.

The Manhattan Project and the Cold War that followed built that instinct into a system: remote sites with controlled access, personnel security investigations, formal classification of information, compartmentalization so that few people saw the whole picture. Most of it worked. The part that did not is the more instructive.

The United States program was penetrated repeatedly. Klaus Fuchs, a theoretical physicist at Los Alamos, passed weapon design information to the Soviet Union. The ring around Julius and Ethel Rosenberg, with David Greenglass and Harry Gold, passed more. Theodore Hall and Saville Sax were identified by the FBI around 1950 through the Venona decrypts of Soviet cable traffic, but never prosecuted, and their names became public only when the decrypts were declassified in 1995 [7]. George Koval, a Soviet military intelligence officer who worked inside the project’s Oak Ridge and Dayton sites, was not publicly identified until Russia awarded him a posthumous Hero of Russia medal in 2007. The Soviet side was penetrated in its turn: Western intelligence ran sources inside the Soviet military and intelligence services for decades, and several were executed after discovery.

The lesson is not that security failed. Fission physics was going to be rediscovered anyway, and the bomb was built in several countries by people who never saw a stolen page. The lesson is that personnel security and counterintelligence are permanent requirements, not wartime measures to be relaxed when the emergency passes. Every institution that handles nuclear material or sensitive information will be a target for as long as it does so, and its defenses have to assume that some of the people inside it are not what they appear to be.

Explore. (a) What does the “copper” story show about scientists’ attitude to information in 1939–42? (b) Two of the Soviet sources were identified from decrypted cables that stayed secret for forty-five years. What does that say about the trade-off between prosecuting a spy and protecting the source that exposed him? (c) Why is the espionage record an argument for permanent personnel security rather than for more secrecy?

4. What actually impedes proliferation

Here is the central claim of the subject, and it is counter-intuitive to most people who come to it fresh: secrecy about how a nuclear weapon works is not the main thing standing between the world and more of them.

The basic principles have been public for a long time. A gun-type weapon fires one sub-critical piece of highly enriched uranium into another; it is simple and robust enough that the design used at Hiroshima was never tested beforehand. An implosion weapon compresses a sub-critical core of plutonium or highly enriched uranium to a super-critical density with precisely shaped conventional explosives; it is more demanding and is the only route that works for plutonium. That much has been in encyclopedias for decades. From 1964 to 1967 the Lawrence Radiation Laboratory at Livermore ran what it called the Nth Country Experiment: three young physicists with no weapons background were asked to produce a credible design using only the open literature. They did, over about two and a half years and three person-years of effort, and the laboratory’s summary report was later released under the Freedom of Information Act [8]. A remark often attributed to J. Robert Oppenheimer holds that the only real secret was that it works.

The consequence has been plain to analysts for decades: the design secrets are long gone, and the binding constraint is access to fissile material — highly enriched uranium (HEU) or plutonium. In 2008 the Washington Post reported that the A. Q. Khan smuggling network had obtained, and may have distributed, blueprints for an advanced weapon [9]. Design information leaks. What cannot be downloaded is a critical mass of the right material.

That is the field’s working consensus, not a law of nature, and it has limits: design information still shortens the engineering path, which is why the Khan network’s customers paid for it (section 6). The ranking below is one of degree, and the case that would reorder it is a proliferator with material in hand and no way to build a deliverable device.

So what stands in the way? Three things, in descending order of importance:

  1. Access to fissile material — highly enriched uranium or plutonium, in quantity, in a usable form. Producing either requires industrial facilities (enrichment plants or reactors plus reprocessing) that are large, expensive, slow, and hard to hide, and that is exactly why the world’s safeguards system concentrates its most intrusive measures on them. The atlas on this site shows where they are.
  2. Engineering capacity to turn a design into a working, deliverable device — a much larger job than the physics, involving materials, precision manufacture, and testing.
  3. Confidence, which has usually meant testing, and a nuclear explosion of any useful size is detectable: the Comprehensive Nuclear-Test-Ban Treaty’s International Monitoring System listens for it worldwide, even though the treaty itself has never entered into force. That barrier is political as much as technical — the moratorium on testing has come under public pressure from more than one weapon state — and its current state, including a disputed allegation of a very small test, is tracked on the situation page.

Safeguards have a limit of their own that matters here. Under a comprehensive safeguards agreement alone, the IAEA can conclude only that the material a state declared stayed in peaceful use. It is the Additional Protocol, adopted after undeclared Iraqi work came to light in 1991, that gives inspectors the access needed to conclude that there is no undeclared material either — the “broader conclusion.” Having the protocol in force is necessary for that conclusion but not sufficient: in the Agency’s most recent annual report, just over half the states with both instruments had received it, and the rest only the narrower one [10]. And verification depends on continuity of knowledge. When safeguarded facilities in Iran were struck in June 2025, and again in 2026, the Agency lost access to the material that had been declared there, including a large stock of highly enriched uranium, and has had none since; by its own account the gap “needs to be addressed with the utmost urgency” [10]. The material barrier, in other words, is only as good as the measurement behind it, and measurement can be lost — to a state that withholds access or to an air strike — faster than material can be made. The current figures are on the situation page.

The implication for how the world defends itself follows directly. Information control has its place, but the primary tools are safeguards, detection, material control and accounting, and intelligence — measuring, tracking, and protecting material, and noticing when it moves. That is engineering and science, and it is the kind of work the rest of this site is about.

Explore. (a) What did the Nth Country Experiment demonstrate, and over what years did it run? (b) Name one situation in which engineering capacity, rather than access to material, would be the binding constraint on a would-be proliferator. What would that imply for where safeguards effort should go? (c) What can the IAEA conclude about a state with a comprehensive safeguards agreement but no Additional Protocol, and what can it not?

5. The treaty and the map of proliferation

The nuclear age began with one state. By the time the Nuclear Non-Proliferation Treaty entered into force on 5 March 1970, five had tested — the United States (1945), the Soviet Union (1949), the United Kingdom (1952), France (1960), and China (1964). The treaty defines a nuclear-weapon state as one that had tested before 1 January 1967, which captures exactly those five; every other party undertakes never to acquire the weapons, and every party — the five included — undertakes to pursue disarmament and to share peaceful technology [11]. France and China stood outside the treaty until 1992. The bargain was preceded by Atoms for Peace (1953), which opened access to nuclear technology, and by the creation of the International Atomic Energy Agency (1957), which would verify its peaceful use.

The treaty has slowed proliferation without stopping it. India tested in 1974 and called the device peaceful; Pakistan tested in 1998; North Korea, having joined and then withdrawn from the treaty, tested in 2006. Israel is a case of its own: it is widely believed to have built weapons around the time the treaty was negotiated, and every public estimate counts it as a possessor, but it has never tested, never confirmed or denied having them, and maintains a deliberate policy of nuclear ambiguity — so the honest statement is that nine states are believed to hold nuclear weapons today, one of them without ever saying so [13]. Against that, the record also holds reversals: South Africa built a small arsenal, dismantled it, and joined the treaty in 1991; Belarus, Kazakhstan, and Ukraine gave up the Soviet weapons on their territory in the 1990s.

The bargain is under strain from both sides. Most parties hold that the disarmament undertaking has not been kept, and the treaty’s review conferences of 2015, 2022, and 2026 all ended without an agreed outcome — the last over language on Iran and on the attacks on its safeguarded facilities [12]. A separate Treaty on the Prohibition of Nuclear Weapons, which the nuclear-armed states reject, has been in force since 2021; its own review cycle is tracked on the situation page.

The Cold War also produced close calls and accidents that belong in any honest account — the Cuban missile crisis of 1962; the Able Archer exercise of 1983; weapons lost or broken at Tybee Island in 1958, Goldsboro in 1961, Palomares in 1966, and Damascus, Arkansas, in 1980 — each a reminder that surety is a live problem for the states that possess these weapons, not a solved one.

The shape of today’s arsenals, in structural terms: Russia and the United States hold the great majority of the world’s warheads; China’s arsenal is growing faster than any other [13]; the other six possessors hold arsenals in the tens to low hundreds. The two independent-looking counts a reader will meet — SIPRI’s yearbook chapter and the Bulletin of the Atomic Scientists’ Nuclear Notebook — are compiled by the same team at the Federation of American Scientists, so agreement between them is one estimate stated twice, not two estimates that agree [13]. For three decades after 1991 the global total fell as the two largest states dismantled retired weapons. That total is still falling, but only because of that dismantlement — the number of warheads in military stockpiles, the ones available for use, began rising again in the 2020s, and SIPRI expects that rise to continue and probably accelerate [13]. The last treaty limiting United States and Russian strategic forces, New START, expired on 5 February 2026 with no successor; its on-site inspections had already stopped in 2020 and its data exchanges in 2023, so the two largest arsenals are now unverified as well as unlimited [14]. Since that date no treaty anywhere limits the size of any state’s nuclear arsenal: China, France, and the United Kingdom were never bound by numerical limits, and the NPT’s disarmament article sets a direction, not a ceiling. The current figures, updated as they change, are on the situation page.

Explore. (a) Which five states does the NPT recognize as nuclear-weapon states, and what is the treaty’s actual criterion? (b) Name four states that acquired weapons after 1970 and two that gave them up. (c) The global warhead total is falling while military stockpiles are rising. Explain how both can be true at once.

6. Proliferation as a business, and the non-state threat

Two developments since the Cold War changed the shape of the problem.

The first was the discovery that proliferation could be run as a business. Abdul Qadeer Khan, the metallurgist regarded as the father of Pakistan’s enrichment program, confessed in 2004 to having run for years an international network that sold centrifuge designs, components, and weapon-related information to Libya, Iran, and North Korea [15]. The network bought parts from suppliers on three continents, shipped through intermediaries, and answered to no government’s policy. It was a new kind of threat — trans-national, commercial, and deniable — and the response to it is why export controls, supply-chain due diligence, and personnel security now matter to a nuclear engineer as much as reactor physics does.

The second was the arrival of non-state actors with declared nuclear ambitions. Al Qaeda tried to buy uranium in Sudan in 1993–94 and was sold a fake [16]; Osama bin Laden described acquiring weapons of mass destruction as a religious duty in a December 1998 interview; a May 2003 fatwa from a Saudi cleric asserted their use was permissible [17]. When the Islamic State seized Mosul in 2014, cobalt-60 teletherapy sources at the city’s hospital and university facilities passed outside anyone’s control for about three years, until the city was retaken in 2017 [18]. What such a group would most plausibly build is not a sophisticated weapon but one of two cruder things: an improvised nuclear device — most likely gun-type, which requires highly enriched uranium — or a radiological dispersal device, “a conventional explosive, such as dynamite, with radioactive material,” whose effect is contamination, disruption, and fear rather than blast [19].

The problem is hard for reasons that have nothing to do with physics. A group of that kind has a small signature, uses unconventional delivery, and cannot be deterred the way a state can. The most reliable strategy against it is denial of material — fissile material for a nuclear device, which brings the argument back to section 4, and radioactive sources for a dirty bomb — working alongside the export-control, intelligence, and personnel-security measures just described.

The world has built a legal framework around that strategy: the Convention on the Physical Protection of Nuclear Material (in force 1987) and its 2005 Amendment (in force 2016), which extends it to facilities and to material in domestic use, storage, and transport; the International Convention for the Suppression of Acts of Nuclear Terrorism (in force 2007); and UN Security Council Resolution 1540 (2004), which binds every state to keep nuclear, chemical, and biological weapons, their means of delivery, and related materials out of non-state hands [20]. The four Nuclear Security Summits of 2010–2016 turned much of that into national commitments.

And material has reached the market. The IAEA’s Incident and Trafficking Database, a voluntary reporting system in which most states participate, has recorded several thousand incidents of nuclear and other radioactive material out of regulatory control since 1993 [21]. About two-thirds are judged unconnected with trafficking — small quantities, lost sources, contaminated scrap; a minority involve trafficking and intent; a handful involve weapons-usable material. One pattern in the record is worth knowing: more than half of the thefts and attempted thefts happened while the material was in transport, and the share has been rising [21]. The current counts are on the situation page. Highly enriched uranium was seized in Prague in 1994 (2.7 kilograms, the largest case outside Russia in the published record), in Germany and Bulgaria in the 1990s, and in Georgia in 2003, 2006, and 2010; in 2011 a group in Moldova offered a kilogram of it for sale and produced a few grams [22]. Most of the quantities were small, and in every case the buyers’ capability fell far short of their intent. What the record demonstrates is that the intent exists and that material has moved; it does not show that anyone has come close to a weapon.

One more fact places the problem on the map. Kazakhstan alone produced 39 percent of the world’s mined uranium in 2024, and with Uzbekistan the two Central Asian producers supplied about 45 percent [23]. The front end of the fuel cycle is global, and so, necessarily, is its security.

Explore. (a) What made the A. Q. Khan network a new kind of threat? (b) Why is an improvised nuclear device more likely to be gun-type than implosion-type? (c) What does the ITDB record show, and what does it not show? Consider what it means that reporting to it is voluntary.

7. The terrain ahead

Several trends are converging on the field at once, and each of them creates work.

New nuclear. Reconstituting and expanding fuel-cycle capacity after decades of atrophy; new countries entering nuclear power for the first time; and a generation of reactor designs whose operating paradigms differ from the large light-water reactor the safeguards system grew up around. Molten-salt reactors circulate fuel as a liquid; pebble-bed reactors handle fuel as hundreds of thousands of graphite pebbles, each holding thousands of TRISO (tristructural-isotropic) coated fuel particles, that circulate continuously through the core in numbers that make item-by-item accounting impractical; microreactors are small enough to move, and the first demonstration units reached criticality in 2026 under Department of Energy authorization rather than a Nuclear Regulatory Commission license, which raises its own question of whose security rules apply [24]. Each is harder to account for by the methods that work for pellets in metal tubes, and each needs new instruments and new approaches.

The HALEU economy. Many of those designs use fuel enriched to just under 20 percent uranium-235 — high-assay low-enriched uranium — rather than the 3 to 5 percent of today’s fleet. HALEU is harder and more expensive to make, comes in a wider variety of forms, and is a more attractive feedstock for a state that intended to enrich further. It changes the economics of enrichment and the calculus of safeguards at the same time.

Fusion. Fusion is a serious engineering enterprise again, and it raises questions the safeguards system has not had to answer: where the tritium for start-up comes from, how lithium-6 breeding is managed, and whether an intense neutron source could be misused and how that would be detected. The United States is moving to regulate fusion machines as byproduct-material users rather than as reactors — the NRC proposed such a rule in February 2026 — a path that would leave the security questions to be decided case by case [25]; whether and when the rule becomes final is tracked on the situation page.

Artificial intelligence and autonomy. AI lowers technical barriers to design and analysis on both sides; deepfakes and disinformation complicate the attribution of any event; and the question of AI in nuclear command and control raises new stability risks — enough that the UN General Assembly adopted its first resolution on the subject in December 2025, demanding that human control and oversight be maintained — by a divided vote in which none of the nine nuclear-armed states voted in favor [26]. Low-cost uncrewed systems — drones — are now a proven threat to nuclear facilities themselves: in February 2025 a drone struck the New Safe Confinement over the destroyed Chernobyl reactor, opened a hole in it, and left the structure unable to perform its confinement function [27]. That was a military strike in a war zone, and how far the design basis of a peacetime facility should extend toward such threats is a live question for every regulator. Space, commercial and military, is a new domain for both threats and tools.

States as attackers. The IAEA definition in section 2 covers any intentional act directed at a facility, but the physical-protection systems built under it were designed against criminals, terrorists, and insiders. The 2020s added a category those systems were not built for: armed attack on civil nuclear facilities by a state — the occupation of the Zaporizhzhia plant in Ukraine since 2022 [12], the Chernobyl strike [27], and the strikes on safeguarded facilities in Iran in June 2025 and again in 2026 [10]. Physical protection was never designed to stop an air force; the law of armed conflict, not the nuclear-security conventions, is what governs such attacks; and whether declaring a facility to safeguards now also marks it as a target is a question the regime has not answered. The module does not adjudicate the attacks themselves; the situation page records who carried them out.

Great-power competition. Strategic stability was bipolar for most of the nuclear age. It is not now. China’s arsenal is growing faster than any other; several regional powers hold weapons; the arms-control architecture is eroding; verification tools are under strain. This is the world in which the next generation of engineers, scientists, and analysts will work.

What all of that demands can be said briefly. New safeguards and security methods and instruments have to be developed and proved, preferably with industry rather than at arm’s length from it. Fast, disruptive development has to be married to the deliberate, stable operation that the “three S’s” — safety, security, safeguards — require. And the work cannot live in stovepipes: research, regulation, and operations have to keep current with one another across a broader set of challenges than the field has faced before. It is a sprint that turns into a marathon, and it needs both capability and continuity in the people who run it.

Explore. (a) Why is a pebble-bed reactor harder to safeguard than a light-water reactor? (b) What makes HALEU a security question as well as an economic one? (c) Apply section 2’s test — who is on the other side of it? — to a state attack on a civil nuclear plant. Which of the four words fits, and what does the answer leave out?

8. Who does this work

Nuclear security is done by engineers, scientists, and policy analysts in roughly six kinds of place. In the United States: the Department of Energy’s National Nuclear Security Administration and its national laboratories and production sites; the Nuclear Regulatory Commission; and the defense agencies and services. Internationally and everywhere: the International Atomic Energy Agency and the wider safeguards community, including regional systems such as Euratom and the Argentine–Brazilian agency ABACC and national regulators in every nuclear state; industry — utilities, reactor vendors, and fuel-cycle companies; and the policy, intelligence, and analytic institutions that need technically trained people. The roles carry names like safeguards engineer, forensic analyst, detection-system designer, fuel-cycle specialist, resilient-design engineer, and arms-control analyst.

Nuclear forensics is worth a paragraph of its own because it answers the question people ask first after any incident: who made this material? The answer comes from analytical chemistry — gamma spectroscopy, mass spectrometry, actinide chemistry in hot cells, metallurgical examination — combined with reactor physics and fuel-cycle knowledge to read a material’s history from its isotopic and chemical signatures. It is a field where the skills of a nuclear engineering degree map directly onto the work.

The requirement the author would put first — a judgment, not a finding: the security side of the nuclear enterprise needs people who understand both the physics, chemistry, and engineering of reactors, fuel, and materials, and the policy, legal, and human dimensions of security. Neither half is sufficient alone.

Explore. (a) Which agency verifies peaceful use internationally, and which regulates civil nuclear facilities in the United States? (b) What question does nuclear forensics answer, and with what tools? (c) Take one of the trends in section 7 and say what a purely technical specialist, and a purely policy specialist, would each miss about it.


Sources

  1. International Atomic Energy Agency, IAEA Nuclear Safety and Security Glossary, 2022 (Interim) Edition, entry “nuclear security”; the same definition as Objective and Essential Elements of a State’s Nuclear Security Regime, Nuclear Security Series No. 20 (2013), para. 1.1.
  2. International Atomic Energy Agency, IAEA Nuclear Safety and Security Glossary, 2022 (Interim) Edition, entry “(nuclear) safety”. Language usage from the IAEA’s French, Spanish, Russian, and Arabic editions and titles.
  3. International Atomic Energy Agency, IAEA Safeguards Glossary, 2022 edition.
  4. U.S. Department of Defense, DoD Nuclear Weapons Surety Program, Directive 3150.02 (24 April 2013, incorporating Change 5, 15 July 2022), glossary.
  5. United Nations General Assembly resolutions 2(I) (1 February 1946) and 3190 (XXVIII) (18 December 1973), on the official languages of the General Assembly.
  6. G. T. Seaborg, “Silver, copper, and honest-to-god copper,” Journal of Nuclear Materials Management 18(1), 16–17 (1989).
  7. Smithsonian Magazine, “The Spies Who Spilled Atomic Bomb Secrets”; National Security Agency, Venona declassified releases (1995).
  8. W. J. Frank (ed.), Summary Report of the Nth Country Experiment, UCRL-50249, Lawrence Radiation Laboratory, Livermore (March 1967), released under the Freedom of Information Act; National Security Archive, “Nuclear Proliferation and the Nth Country Experiment,” 23 January 2025.
  9. J. Warrick, “Smugglers Had Design for Advanced Warhead,” Washington Post, 15 June 2008.
  10. International Atomic Energy Agency, “IAEA Draws Safeguards Conclusions for 179 States,” Safeguards Implementation Report for 2025, 15 July 2026; IAEA report to the Board of Governors on Iran, September 2026, and the Board’s resolution of 9 September 2026, as reported by RFE/RL, 2 and 9 September 2026.
  11. Treaty on the Non-Proliferation of Nuclear Weapons, opened for signature 1 July 1968, entered into force 5 March 1970, Articles IV, VI, and IX.3 (UN Office for Disarmament Affairs); Arms Control Association, “Timeline of the NPT.”
  12. Arms Control Association, “2026 NPT Review Conference Stymied by Disputes,” Arms Control Today, June 2026.
  13. Stockholm International Peace Research Institute, SIPRI Yearbook 2026, chapter 8, “World nuclear forces” (H. M. Kristensen and M. Korda; figures as of January 2026); H. M. Kristensen, M. Korda, E. Johns and M. Knight-Boyle, “Status of World Nuclear Forces,” Federation of American Scientists, as of early 2026, and the Bulletin of the Atomic Scientists Nuclear Notebook series by the same authors (country columns 2025–2026).
  14. Arms Control Association, “New START Expires; U.S. Urges ‘Modernized’ Treaty,” Arms Control Today, March 2026.
  15. International Institute for Strategic Studies, Nuclear Black Markets: Pakistan, A. Q. Khan and the Rise of Proliferation Networks, Strategic Dossier (2007).
  16. National Commission on Terrorist Attacks Upon the United States, The 9/11 Commission Report (2004), chapter 2.
  17. R. Mowatt-Larssen, “Al Qaeda’s Pursuit of Weapons of Mass Destruction,” Foreign Policy, 25 January 2010 (timeline).
  18. Institute for Science and International Security, “Cobalt-60 Sources in Mosul: Recovery and Lessons for the Future,” July 2017.
  19. U.S. Nuclear Regulatory Commission, Backgrounder on Dirty Bombs.
  20. Nuclear Threat Initiative, treaty summaries for the Convention on the Physical Protection of Nuclear Material and its 2005 Amendment, and for the International Convention for the Suppression of Acts of Nuclear Terrorism; United Nations, 1540 Committee fact sheet on Security Council resolution 1540 (2004).
  21. International Atomic Energy Agency, “IAEA Database: About 55% of Nuclear and Other Radioactive Material Thefts Since 1993 Occurred During Transport,” press release, 23 March 2026; as reported by American Nuclear Society, Nuclear Newswire, 25 March 2026.
  22. Center for Public Integrity, “The fuel for a nuclear bomb is in the hands of an unknown black marketeer from Russia, U.S. officials say,” table of HEU and plutonium seizures 1992–2011; Voice of America report on the April 2016 Georgian arrests (the material was depleted uranium).
  23. World Nuclear Association, “World Uranium Mining Production,” updated 21 September 2026.
  24. U.S. Department of Energy, “U.S. Department of Energy Meets President Trump’s Goal, Delivers Third Advanced Reactor,” 1 July 2026 (the third of the demonstration reactors authorized under the Reactor Pilot Program and the Nuclear Energy Launch Pad to reach criticality in 2026).
  25. U.S. Nuclear Regulatory Commission, “Regulatory Framework for Fusion Machines,” proposed rule, 91 Federal Register, 26 February 2026, Docket NRC-2023-0071.
  26. United Nations General Assembly, resolution 80/23, “Possible risks of the integration of artificial intelligence into command, control and communications systems of nuclear weapons,” adopted 1 December 2025.
  27. American Nuclear Society, “IAEA: Chernobyl drone strike latest threat to nuclear safety in Ukraine,” Nuclear Newswire, 18 February 2025; NucNet, “IAEA says Chernobyl shelter lost primary safety functions following drone strike,” 8 December 2025.

This material is based in part upon work supported by the Department of Energy National Nuclear Security Administration through Defense Nuclear Nonproliferation’s Enabling Capabilities in Technology Consortium under Award Number DE-NA0004197. The views and opinions expressed are the author’s own and do not necessarily state or reflect those of the United States Government or any agency thereof.

Last reviewed 2026-09-22.