North Korea and the Proof of Nuclear Adherence

September 2021
By Ankit Panda and Toby Dalton

In May 2021, the Biden administration announced its intention to pursue a “calibrated, practical approach that is open to and will explore diplomacy” with North Korea. The intention was to distance its approach from those of President Joe Biden’s two immediate predecessors. As White House spokesperson Jen Psaki emphasized, “[O]ur policy will not focus on achieving a grand bargain, nor will it rely on strategic patience.”1

In a subtle but potentially transformative decision, the administration also signaled it would seek “practical progress” with North Korea in ways that could increase “the security of the United States, our allies, and deployed forces.” This statement acknowledges the obvious: that U.S. and allied interests could be served by measures that fall short of the complete denuclearization of the Korean peninsula, which remains a long-term objective.

Although the administration does not use the phrase “arms control” in describing its North Korea policy, achieving any “practical progress” would require limiting the quantitative growth and qualitative improvement of North Korea’s nuclear capabilities. Among the many difficult issues that U.S. negotiators would have to address with Pyongyang, if and when negotiations resume, is how North Korean compliance with such limits could be verified and monitored.

Practical Verification for Practical Progress

In the past, verification has proved a source of considerable tension when implementing nuclear agreements with North Korea. Under the 1994 Agreed Framework, long before North Korea gained the range of nuclear capabilities it has, U.S. inspectors successfully obtained on-site access to suspected sites2 but only after protracted negotiation with North Korean officials. The North’s fundamental mistrust of the United States, other major powers, international organizations, and the entire, highly intrusive verification process complicated these efforts.3 In late 2008, Pyongyang’s misgivings about a verification protocol to the 2007 six-party talks4 gave way to a slow-simmering crisis that eventually boiled over, resulting in the expulsion of International Atomic Energy Agency (IAEA) monitors in April 2009. A month later, North Korea carried out its second nuclear test.5

Despite this, Pyongyang understands that verification and monitoring are a sine qua non of any potential nuclear agreement. For instance, after the Trump administration rejected North Korea’s proposed concessions at the February 2019 summit in Hanoi, North Korean Foreign Minister Ri Yong Ho emphasized that Pyongyang’s offer to dismantle “nuclear material production facilities in the Yongbyon area”6 entailed doing so “in the presence of U.S. experts.” The Pyongyang Declaration of September 2018,7 agreed between South Korean President Moon Jae-in and North Korean leader Kim Jong Un, similarly included a clause whereby the North would dismantle its Dongchang-ri (or Sohae/Yunsong)liquid-propellant engine test stand “under the observation of experts from relevant countries.”8Fuel Rod Fabrication Plant and other facilities at North Korea's Yongbyon nuclear complex in 2007.  (Image: Google Earth, © 2021 Maxar Technologies)

By 2021, the Fuel Rod Fabrication Plant had been renovated and expanded. Nuclear expert Olli Heinonen wrote for the 38North website that commercial satellite imagery shows that since 2009

Now that Biden administration officials have unambiguously stated9 that denuclearization is a “long-term objective” rather than the singular and immediate goal demanded by President Donald Trump, U.S. policymakers should begin working on creative, practical, perhaps even unorthodox approaches to verification and monitoring.

The U.S. government has considerable experience and expertise in monitoring and verifying nuclear and missile restraints. Yet, the shift in assumptions implied by the Biden policy, from a one-shot denuclearization agreement to incremental steps, means that verification experts will confront unprecedented challenges that will require new tools and approaches. If the administration does not plan for these situations now, negotiators may be inhibited in the types of progress they may be able to clinch when negotiations resume.

North Korea is likely to reject orthodox and invasive verification measures, such as “anytime, anywhere inspection,” at least at the beginning of a denuclearization process. With an ever-growing arsenal that includes an estimated 20 to 60 nuclear warheads and an expansive inventory of nuclear delivery systems, Pyongyang’s negotiating leverage is far greater than it has been in the past. As a result, “practical progress” is highly unlikely to begin with the return of IAEA inspectors to the Yongbyon nuclear complex or, more ambitiously, with on-site inspections at missile operating bases that North Korea has refused to acknowledge.

Even so, negotiators can and should seek to maximize the verifiability of any potential agreements. That will be critical to the political viability of these agreements and will ensure that progress toward denuclearization is observable and measurable. In this way, verification and monitoring will serve as a means toward the ultimate goal of denuclearization, rather than an end in themselves.

Novel Approaches to Verification and Monitoring

As the Biden administration began its North Korea policy review, the Carnegie Endowment for International Peace convened a group of international technical experts to study potential new approaches to verification and monitoring. Part of the group analyzed novel approaches to verifying declared items in a potential future agreement, such as missiles, missile launchers, fissile material, and, eventually, nuclear warheads. Others explored conceptual, technical, and methodological approaches to building a layered monitoring system. This included an examination of probabilistic verification and compliance assessment, the applicability of open-source intelligence tools, and the promise of a nodal monitoring system.

The group also considered approaches to an export-import regime that might limit North Korea’s ability to procure critical goods for its weapons of mass destruction programs or to sell such items to third countries. Finally, despite the considerable differences between the cases of North Korea and Iran, the group assessed the applicability of some of the innovative verification and monitoring provisions that were included in the 2015 Joint Comprehensive Plan of Action with Iran.10

These studies offer some suggestions for how the Biden administration could seek to verify and monitor prospective practical agreements with North Korea, such as a missile freeze, fissile material controls, and limitations on deployed missiles.

Missile Test Freezes and Beyond

North Korea has made important qualitative strides in its missile capabilities in recent years, particularly under Kim, and now possesses several types of missiles assessed to be nuclear capable. There is some precedence for Pyongyang to agreeing to negotiated, albeit temporary, missile restraints. With the goal of supporting then-ongoing diplomacy, the Berlin agreement in September 1999 formalized a nearly seven-year-long freeze on long-range missile tests by North Korea. The Leap Day deal in February 2012 established a moratorium on long-range missile launches, although it collapsed in less than two months over differences between the United States and North Korea on whether it covered space launches. Most recently, during the Trump administration’s diplomatic outreach with North Korea in 2018 and 2019, Kim voluntarily announced a moratorium on long-range missile testing, which has since been rescinded.

By 2021, the Fuel Rod Fabrication Plant had been renovated and expanded. Nuclear expert Olli Heinonen wrote for the 38North website that commercial satellite imagery shows that since 2009 Pyongyang has never submitted to intrusive verification and monitoring of its missile capabilities and missile-related industrial complex, so whatever the form of future negotiations, this issue is sure to be contentious. Even so, missile freeze agreements could take various forms, with implications for how they might be verified.

By any measure, a test freeze remains the easiest objective for negotiators. It could cover the flight testing of fully assembled missile systems and the testing of certain subsystems, including static ground testing of rocket boosters. Both types can be verified remotely through the use of U.S. space-based infrared sensors, which are optimized for the detection of the hot plumes associated with missile launches and ground testing. North Korean testing of nuclear-capable cruise missiles, however, may present certain problems for space-based monitoring. Cruise missiles that have been tested are not known to be nuclear capable, but Kim has indicated that a new intermediate-range cruise missile under development may be nuclear capable.

The value of a test freeze decreases as North Korea generally becomes more sophisticated and experienced with missile technologies. Under a test freeze, for instance, there would be no restraints on North Korea producing more missiles of types that have already been proven or attempting other qualitative upgrades to existing missiles, such as improved guidance.

Because of this limitation, policymakers may choose to seek two considerably more ambitious objectives: a freeze on missile production or on missile deployments. Neither has any precedent with North Korea, but each would represent marked progress toward subsequent denuclearization steps. A production freeze would cap the growth of North Korea’s nuclear force. A deployment freeze would limit and perhaps, over time, degrade the size and readiness of North Korea’s deployed arsenal. Both freezes could be observed with some confidence via remote sensing capabilities, but verification and monitoring would benefit considerably from on-site inspections or other types of intrusive on-the-ground monitoring.

Although freezes on production and deployment would be infinitely preferable, a freeze on testing, implemented properly, could facilitate the conditions for further diplomatic progress with North Korea and open the door to the more complex verification arrangements required to support production and deployment freezes.

Applying Flexible Safeguards

Traditionally, U.S. and international efforts to restrain and reverse North Korea’s nuclear ambitions have focused on freezing the production of weapons-grade fissile material, as in the 1992 North-South Joint Declaration on Denuclearization of the Korean Peninsula, the 1994 Agreed Framework, and the agreements of 2005 and 2007 stemming from the six-party talks. Although the IAEA monitored the freezes, its activities fell well short of applying traditional IAEA safeguards to relevant facilities at the Yongbyon complex.

IAEA safeguards remain the standard for verifying that nuclear materials are not diverted to weapons use. Yet, future attempts to monitor and safeguard North Korean facilities at Yongbyon and elsewhere need not begin with demands for complete access to these facilities, records, and, most critically, nuclear material. Instead, utilizing traditional safeguards tools through more flexible and gradual approaches to monitoring negotiated fissile material controls would initially make sense.

For example, rather than insisting that North Korea present a complete declaration of nuclear materials to be verified, a more practical agreement could stipulate a piecemeal approach involving the monitoring of specific facilities, such as the uranium-enrichment hall at Yongbyon; of materials, such as separated plutonium, not in weapons form; or stages of the fuel cycle, such as uranium conversion. Safeguarding declared waste materials, such as spent fuel, could also serve as a useful, early stepping-stone to more comprehensive safeguards.

Meanwhile, monitoring could initially be focused on verifying the nonoperational status of facilities or the presence of specific materials. As North Korea complied with such limited safeguards activity, more intrusive access could become feasible. A flexible approach along these lines would avoid the problem of placing high hurdles, such as a demand for a complete declaration or full access, too early in a negotiation, yet would still permit the IAEA to begin to assemble a more complete picture of the North Korean nuclear enterprise.

Broken seals that had been used to tag nuclear equipment under International Atomic Energy Agency safeguards. North Korea ceased cooperating with the IAEA in April 2009 and international monitors withdrew after removing all IAEA seals and switching off surveillance cameras. (Photo by IAEA)This piecemeal approach to safeguards need not preclude the IAEA’s eventual return to traditional verification and monitoring activities in North Korea. To the contrary, such an approach would be central to the final goal of complete denuclearization. Despite its lack of a presence in North Korea since 2009, the IAEA has continued to use open sources and a variety of analytical techniques to maintain its readiness for an eventual return to the country.

Monitoring Missile Bases

Introducing restraints on North Korea’s existing nuclear weapons would be an important means of reducing the threat to the security of the United States and its allies. No prior agreements with North Korea dealt with nuclear weapons as such, only with fissile materials and missiles. Pyongyang seems especially unlikely to permit inspections or other means of intrusive transparency at nuclear missile bases, none of which it has even declared, so how can meaningful restraints be implemented and monitored?

One approach could be to confine North Korean missiles to certain declared bases and monitor the perimeters of those bases to ensure that missiles do not leave. This type of restraint would not diminish the quantitative threat to the United States, but it could address it qualitatively by affecting North Korea’s nuclear posture and the readiness of its nuclear forces.

For this type of restraint, a nodal monitoring system could prove useful. Similar to perimeter portal monitors used in other nuclear security and arms control applications but making use of advanced sensor and network technologies, such a system would monitor the movement of specified items. Nodes consisting of a variety of sensors would be placed at specific ingress and egress points. This would require inspectors to have physical access to the perimeter of such a facility but not necessarily a persistent on-site presence. In a case where on-site presence might be tolerated by North Korea, for instance, at a well-known and declared complex such as Yongbyon, additional nodes could be arrayed in order to confine specific items to an individual building.

Although this approach could make verification more technologically complex, it would also be more flexible. It may be initially more acceptable to North Korea than intrusive inspections. With time and deeper implementation of restraints, Pyongyang may allow for the progressive expansion of nodal monitoring at certain sites.

The technology base for nodal monitoring is well developed, but building a robust networked system, especially one that could be left unattended, would require additional research, development, and testing. Beyond providing for multiple types of sensors, a ready-to-deploy node should include a secure, reliable, and encrypted data relay, in situ power generation, and the ability to operate without excessive maintenance. To assuage Pyongyang’s concerns about a nodal system facilitating unsanctioned intelligence gathering and to build trust, North Korean technical experts could participate in the testing and development of these nodes, subject to export control restrictions.

New Approaches and Tools

As the above examples indicate, verifying and monitoring the various elements of North Korea’s increasingly vast nuclear weapons and missile enterprise will require new approaches. Even a modestly successful process of denuclearization will require policymakers to cope with an expanding array of North Korean facilities, materials, and processes. Obstacles are likely to emerge as verification grows more complex and Pyongyang remains reluctant to permit international access. Meanwhile, political concerns about North Korean secrecy and noncompliance are certain to persist under any agreement. To make practical progress, policymakers must avoid letting perfect verification become an enemy of sufficient verification.

One way policymakers could address these issues would be to adopt a framework of probabilistic verification. In situations where high-confidence monitoring of the few key facilities or activities of interest is unavailable or infeasible, which is manifestly the case in North Korea, negotiators could instead seek to verify compliance by monitoring a broader range of facilities and activities with lower confidence levels. As long as enough of North Korea’s total nuclear and missile complex is covered by such an agreement, Pyongyang would be unlikely to gain a meaningful advantage even if it is able to successfully evade monitoring at one single facility.11

Similarly, to support verification, negotiators could rely more heavily on open-source intelligence. An often-overlooked set of tools with growing contemporary relevance, it involves information that is not derived from classified sources and is already a major source of analytical insight into North Korean activities among civil society organizations and journalists. Although governments have traditionally favored national technical means and other proprietary tools, the use of open sources in the North Korean context could usefully augment verification and monitoring, especially in the case of a limited agreement that may be likely to emerge early in a longer negotiating process.

By their very nature, open sources are shareable, which can help build confidence in compliance among parties to an agreement. This could also allow for reciprocity, whereby North Korea, despite lacking any national remote sensing capabilities, could use commercially available satellite imagery to verify, for example, certain types of U.S. military activities covered under an agreement. Although open-source intelligence methods may fall short of national technical means, they can complement proprietary verification tools and more readily be shared with the international community, especially when some parties to an agreement, such as Beijing and Washington, are unlikely to cooperate with each other. In particular, as commercial open-source sensors broaden to include thermal, near-infrared, and other nonvisible spectrum sensors, their role in verification can grow. Several North Korean facilities of interest to negotiators, including missile test stands, the Punggye-ri nuclear test site, and heavy-military-vehicle manufacturing plants, can already be usefully monitored with commercially available sensors.

Planning Ahead

At the moment, diplomatic momentum for any negotiated agreement with North Korea remains low. The United States and North Korea have had no meaningful bilateral interactions since the fizzled 2019 Hanoi summit between Trump and Kim. A working-level meeting in Stockholm in October 2019 was terminated almost immediately by the North Korean side, just days after Pyongyang carried out the inaugural launch of the Pukguksong-3 submarine-launched ballistic missile, making clear its disinterest in diplomacy.

Moreover, weeks before North Korea locked down its borders in January 2020, citing a threat to its “national survival” from the emerging COVID-19 virus, Kim disavowed his earlier April 2018 self-imposed moratorium on long-range missile and nuclear testing in an address to the Workers’ Party of Korea plenum. With the apparent end of the diplomatic charm offensive that began with Kim’s outreach to South Korea ahead of the Pyeongchang Winter Olympic Games in early 2018, North Korea’s self-restraint has waned. Two military parades, in October 2020 and January 2021, further exhibited the fruits of Pyongyang’s accelerated nuclear force modernization.

Against this backdrop, the task of verifying and monitoring prospective agreements to restrain North Korea’s nuclear and missile capabilities presents no shortage of challenges. Innovative and even unorthodox approaches and tools can help render these challenges more manageable. With realistic expectations about what is feasible given persistent mistrust between North Korea and the outside world, the Biden administration, along with allies South Korea and Japan and other international partners, could meaningfully realize its objective of near-term threat reduction. Traditional verification remains the preferred standard, but practical progress in the near term will require novel methods for verification and monitoring in North Korea.



1. “Press Gaggle by Press Secretary Jen Psaki Aboard Air Force One en Route Philadelphia, PA,” The White House, April 30, 2021,

2. Howard Diamond, “U.S. Says N. Korea Site Nuclear Free; Perry Visits Pyongyang,” Arms Control Today, April 1999,

3. Joel Wit, “What I Learned Leading America’s 1st Nuclear Inspection in North Korea,” NPR, January 22, 2019,

4. Peter Crail, “Six-Party Talks Stall Over Sampling,” Arms Control Today, January 2009,

5. Choe Sang-hun, “North Korea Claims to Conduct 2nd Nuclear Test,” The New York Times, May 24, 2009.

6. North Korean Foreign Minister Ri Yong Ho, press conference, Hanoi, February 28, 2019,

7. “Pyongyang Joint Declaration of September 2018,” The National Committee on North Korea, September 19, 2018,

8. This test stand remains in place.

9. Colin Kahl, Keynote address at 2021 Carnegie International Nuclear Policy Conference, June 23, 2021,

10. For summaries of each of these concepts, tools, and approaches to verification and monitoring, see New Approaches to Verifying and Monitoring North Korea’s Nuclear Arsenal, eds. Ankit Panda et al. (Washington: Carnegie Endowment for International Peace, 2021),

11. For a detailed discussion of probabilistic verification in the North Korean context, see Mareena Robinson Snowden, “Probabilistic Verification: A New Concept for Verifying the Denuclearization of North Korea,” Arms Control Today, September 2019,


Ankit Panda is the Stanton Senior Fellow in the Nuclear Policy Program at the Carnegie Endowment for International Peace (CEIP). Toby Dalton is a co-director and senior fellow in the program. With Thomas MacDonald and Megan DuBois, they are editors of New Approaches to Verifying and Monitoring North Korea’s Nuclear Arsenal (2021), on which this article is based.