Introduction: Verification in an Era of Rapid Technological Change

In this CNTR Monitor, we adapt and broaden our definition of verification to reflect a changing security environment in which few arms control treaties with formal verification regimes remain in place and in which geopolitical tensions and shifts in the balance of power call for flexible approaches to both longstanding and emerging security problems. Significantly, in this environment, it is more important than ever to continue producing reliable and internationally accessible knowledge about activities that could undermine peace and security at the regional or global level. This in itself can act as a deterrent or means of dissuasion, establish accountability, help increase confidence in compliance with treaty obligations, detect violations of agreements, and prevent information gaps and reliance on worst-case assumptions in crucial areas. It is precisely this broad understanding of internationally accessible stability- and security-relevant knowledge production aimed at increasing transparency to which we refer in using the term verification. In this perspective, verification can be based on treaties, but does not exclusively rely on them, aims to establish whether states are acting responsibly, and could even include new actors, such as civil society, in such knowledge production.

Such a broad understanding of verification reveals important dynamics and points of friction that are likely to become increasingly relevant in the future. For instance, the balance between transparency and intrusiveness, long a matter of contention, may become even more consequential with advances in fields such as satellite imagery and AI-supported analysis. The relevance and function of negotiated restrictions of data-gathering in formal verification protocols may shift in this area as public access to knowledge becomes increasingly fine-grained. The 2026 CNTR Monitor seeks to illuminate the interplay between these and other technological and political dynamics and offer recommendations how to react to them.

New Technologies for Verification Purposes

Many technologies are currently seeing rapid improvements that could—and in some cases already do—support verification. The most relevant elements include the improved safety of inspectors, data collection and data analysis, attribution methods, and cost reduction.

One area that warrants particular attention in this context is sensor technology. New sensors are becoming increasingly smaller while simultaneously offering higher resolution and greater robustness. They are easier to handle and more durable, for example because they operate at lower voltages. This reduces costs, which is especially important for nuclear safeguards. Particularly in light of the anticipated introduction of small modular reactors and other new reactor technologies, new sensor technologies, especially when deployed remotely, could help the International Atomic Energy Agency (IAEA) maintain a sufficiently dense yet affordable verification network.

Some of these advantages also hold true for future quantum sensing. Although quantum sensing has not (yet) produced any dramatic improvements in terms of resolution or robustness, it could significantly reduce the costs of analysis in some applications, help detect anomalies that trigger verification or investigation activities, and supplement—or, in the longer term, even replace—existing verification methods, for example, in identifying chemical substances or enabling inertial navigation in remote locations.

Advances in satellite imagery also offer important benefits for verification. The resolution of optical, infrared, and radar images is continually improving, although, unlike in other fields, the associated cost trends are hard to predict. While the cost of delivering payloads to orbit per kilogram is falling, higher resolution depends on the surface area of the mirrors used in telescopes, and larger, flawless mirrors are—ceteris paribus—more difficult to manufacture and transport. By fusing satellite images from multiple sources using AI, even when imagery is sub-state-of-the-art quality, it is possible to reveal features that would otherwise remain undetectable even in the highest-quality single-source imagery available. While there is no evidence that AI-enabled data fusion from other sources produces effects comparable with those achieved with satellite data, improvements in data analysis can generally be expected.

Yet, the use of AI in verification mechanisms remains a sensitive issue. The IAEA, for example, already employs machine learning in its analysis of large datasets and satellite imagery, while the Organisation for the Prohibition of Chemical Weapons (OPCW) has recently completed a review of the role of AI in its own work.¹ However, while AI has the potential to support verification, its use also carries risks of misinterpretation and “hallucinations” and raises second-order concerns about trust in AI, particularly with regard to the quality of training data, the potential for manipulation, and the explainability of AI-supported findings.² Measures can and should be taken to address these problems and enhance the credibility of AI-supported activities.

In the biological field, genetic engineering attribution has been recognized as one possible means of supporting and complementing verification and attribution by providing information on the origins of genetic modifications. This in turn could, for example, help trace genetically modified pathogens back to their source, which is important in the event of a deliberate biological incident or a disease outbreak of unknown or ambiguous origin. Assuming that the available data are expanded, streamlined, refined, and adapted to real-world scenarios, genetic engineering attribution might support and complement other verification and attribution methods.

In summary, as the contributions to this Monitor show, many technological developments have the potential to support, enhance, and transform verification. While some may offer only incremental improvements or cost reductions, others—particularly those involving the use of AI to extract information from multiple or diverse data sources—could enable significant advances in generating new insights. As also discussed in this Monitor, seizing these opportunities must go hand in hand with recognizing and addressing potential challenges in applying new technologies to verification, exploring their role beyond traditional settings, and creating an ecosystem in which verification research can thrive.

The Fine Line Between Verification and Spying

In the traditional, narrow definition, verification is typically understood as the essential mechanism for increasing confidence that the parties to a treaty are complying with their obligations and enhancing the likelihood that noncompliant behavior will be detected. The high stakes involved in international security mean that without this provision, there would be a greater risk that a member state might secretly violate the agreement for its own advantage. However, the term “verification” should not be understood as implying a degree of absolute certainty that even the strictest surveillance regime cannot realistically provide. What verification can do, however, is provide a technical basis for assessing compliance, enhance trust, improve the effectiveness of agreements, and increase the political and material costs for noncompliant behavior. In principle, all the technologies described in this Monitor can or could be used to support and enhance verification and reduce uncertainty about the behavior of the parties to an agreement.

However, a number of further aspects require consideration, as overly intrusive verification could ultimately cause more harm than good. States therefore need to carefully balance effectiveness, intrusiveness, and national interests when designing verification mechanisms, a task that requires both technical expertise and political judgment. Importantly, verification generally does not aim to detect small or minor, often unintended, violations, but rather focuses on those actions considered militarily significant. If enhanced verification technologies provide increasingly detailed information on states’ behavior and lead to the more frequent identification of smaller violations, including those resulting from inadequate national implementation of obligations, additional or different procedures may be required to address such issues collectively and cooperatively in order to avoid unnecessary tensions within a regime.

Moreover, the line between permissible verification and intended or unintended industry or military espionage might easily be crossed—either when highly intrusive verification methods collect additional information as a kind of “bycatch,” or when such information jeopardizes core aspects of a verified party’s security or proprietary information while providing only marginal benefits to the other parties. Even the perception of this risk could affect the level of acceptance of and willingness to participate in verification activities.

This situation is further complicated by the fact that some new technologies capable of enhancing verification can also be used unilaterally to gather information, potentially raising concerns about espionage or intrusive surveillance and reconnaissance. In the space domain, for example, autonomous approach functions provide the possibility to obtain extensive information about “inspected” satellites, and includes a potential for sabotage. Recent advances in data fusion may enable states and nongovernmental actors alike to derive information from commercially available imagery that could potentially compromise national security. In the past, verification by national technical means (NTM)—meaning satellite surveillance of nuclear bases—was an important component of verification under the major, now-expired nuclear arms control treaties between the Soviet Union and the United States. These included obligations not to obstruct the other sides’ activities. However, no regime currently exists that explicitly encompasses and regulates such unilateral information gathering. Depending on how they are used, technological advances in verification could therefore contribute to increased surveillance and, consequently, heightened tensions.

Other Approaches to Verification

Based on a definition of verification as a process of knowledge production, some contributions to this year’s Monitor broaden the focus beyond traditional technology- and state-based approaches to encompass other aspects and forms of verification.

For instance, improving national implementation of existing regulations is regarded as one avenue complementary to verification. If adapted for arms control, systems for implementation review (SIRs) could be applied in several arms control fields covered in this Monitor. Rather than monitoring activities and installations, SIRs in arms control could provide forums for collectively reviewing, discussing, and improving national implementation measures, thereby contributing to the effectiveness of arms control regimes and regulations in a different way.

Specific features of outer space and cyberspace make verification particularly challenging. Both are characterized by a high degree of dual use, which creates distinct challenges in each domain. In outer space, the growing autonomy of spacecraft has consequences for verification, both by making it increasingly difficult to determine the nature of activities and by requiring verification activities to remain nonintrusive and retain their character as a form of collective knowledge production rather than unilateral intelligence gathering.

While verification is thus possible in and from outer space in principle, the situation is more complex in cyberspace. Here, the strong dual-use nature of the technology and its ubiquitous application make classical verification highly challenging, though not entirely impossible.³ Drawing on experience with the Chemical Weapons Convention, regulation could be based on intent rather than capabilities and place greater emphasis on attribution than on verification in the traditional sense.

Lastly, the role in verification of nongovernmental actors and experts outside government and intelligence services also deserves consideration. Bolstered by technological advances and the resulting greater accessibility and improved quality of publicly available data, including commercial satellite imagery, open-source intelligence (OSINT) and investigations by nongovernmental actors can contribute to providing additional information on verification-relevant issues and informing public debates and political decision-making, notwithstanding some problems and potential frictions in their application. As diverse as these examples are, they demonstrate that discussions of verification in the new security environment need to move beyond traditional conceptions and consider different approaches in the wider context of arms control verification.

Designing an Ecosystem to Develop Verification Technologies

Most of the advances that promise to significantly enhance verification capabilities originate in basic research. Research in solid-state and atomic physics, optics, and information theory has driven the second quantum revolution, opening up new possibilities for its application to verification. Advances in radiation detection developed for fundamental particle physics have generated new capabilities that can be exploited for nuclear safeguards. Computer science is developing new AI systems and architectures that can be applied to a wide range of verification-relevant tasks. The capabilities of microbial forensics rely on progress in microbiology and genomics. In other words, fundamental advances that later shape verification technologies are typically developed without arms control in mind.

However, adapting and applying the results of frontier science to verification technologies that are ripe for implementation is far from straightforward. Every chapter of this year’s CNTR Monitor Focus provides insights into the intricacies of verification at the intersection between science and politics, their mutual influence, and the ways in which technologies can be used for verification purposes. The distinct security aspects and specific characteristics of the weapons in question must be taken into account. The capabilities and limitations of such systems must be scientifically assessed by scholars with expertise in verification regimes and in the specific items, facilities, and general characteristics of evidence and its environments. Ultimately, these systems must be tested under realistic conditions.

It is in these areas that the authors of the CNTR Monitor identify the most significant obstacles to harnessing the best available science for the most effective verification. The development requirements for verification techniques are highly specific and target a small user base. Sustained financial support is needed to ensure that researchers and producers can continue to undertake these niche tasks. One example of a functional development infrastructure is the Member State Support Programme coordinated by the IAEA in the field of nuclear non-proliferation.⁴ For most other arms control or disarmament domains, such ecosystems are rare or entirely absent.

First and foremost, bringing frontier developments into verification practice requires sustained and structured support for interdisciplinary research environments at universities and other institutions. Here, the applicability of frontier research to the development of novel verification techniques can be assessed, and research aimed at developing such techniques undertaken. This should be followed by development activities to advance verification technologies toward higher levels of technological readiness. Lastly and equally importantly, mission-driven validation in realistic settings, such as field trials, is crucial. Such trials could initially be carried out as standalone tests, but should ultimately be integrated into exercises that combine the application of verification methods with the political processes surrounding verification. Here, collaboration between governmental and nongovernmental actors from academia, applied research, and industry could contribute to the design of realistic scenarios while enhancing the technical feasibility and political acceptance of the verification measures in question.

Conclusion

Dedicating an entire volume to verification in the current geopolitical climate may, at first glance, seem counterintuitive. After all, many arms control agreements containing verification provisions have either expired or become dysfunctional, while traditional verification approaches often seem ill-equipped to deal with new political realities and technological developments. Yet, some verification systems—such as those maintained by the IAEA or the OPCW—remain in place and may be more necessary than ever. These systems could benefit greatly from harnessing technological advances and coupling them with existing verification methods, where feasible and appropriate.

Despite these challenges, there are still developments that give cause for hope. In the biological field, where no verification system is currently in place but—contrary to the broader trend in arms control—political agreement exists to at least discuss verification options, technological assessments of potential verification methods represent important contributions to both the political and technical debate.

Although such debate remains the exception, and despite the absence of any signs that the security situation in Europe and worldwide is easing, it is important not to lose sight of the possibilities for verification. After all, if the broader political climate improves, solutions should be ready for the arms control challenges that will inevitably arise. Monitoring technological developments and exploring how they might be harnessed for arms control, disarmament, and non-proliferation purposes is therefore essential. This Monitor offers ideas for putting new technologies to constructive use in these areas.