The Sweden-based field demonstration, held in June 2025 within the NATO Science and Technology Organisation (STO), opened the door to a new era in the detection of CBRNe (Chemical, Biological, Radiological, Nuclear and Explosive) threats.[i] This five-day event, conducted at the Swedish Defence Research Agency’s Grindsjön Facility, included operational testing of portable sensors for trace-level detection of explosive and pharmaceutical materials. Coordinated by two separate technical panels within the NATO STO, this demonstration (SET-316 and SET-HFM-324) not only evaluated sensor performance but also tested the feasibility of next-generation detection architectures for use hybrid threat environments where the distinction between military and civilian use is blurred.
These sensor systems, developed to detect surface contamination, have been tested in both commercial and prototype formats. The ability to detect trace levels of substances generally considered harmless, such as pharmaceutical residues, but which have the potential to be used in hybrid threats, is directly related to NATO’s goal of enhancing situational awareness. In this context, the demonstration is not only a technical demonstration but also a preliminary field experiment that explores military ethics, rules of engagement (ROE), and dual-use. This technological leap is considered a critical milestone that opens up discussions about the legal and ethical limitations of AI enabled sensors deployed in civilian areas in the future.
It appears possible that AI based CBRNe systems could operate with limited autonomy. In this case, when systems detect hazardous chemicals or trace explosives, final decision-making authority would remain with humans, while preliminary assessment and threat analysis could be handled by AI. Such a model could increase operational speed and efficiency while operating in full compliance with NATO’s Rules of Engagement (ROE). However, the field applicability of this approach may vary depending on the system’s reliability coefficient and the likelihood of generating false positives.
The definition of “algorithmic response” within scope of ROE may be considered. Expanding engagement protocols used within NATO to encompass not only human centred decision-making mechanisms but also AI-supported pre-alert and response call systems may become a technical and legal imperative. Especially in situations requiring immediate response, such as CBRNE threats, AI’s ability to determine the threat level and provide this data to the operator could lead to a new engagement model. This could necessitate the restructuring of NATO’s operational procedures and training programs.
It’s possible that new protocols specific to CBRNe systems will be developed in international law. The use of AI-enabled sensors, particularly in civilian areas, may necessitate new regulations under both the Geneva Conventions and international human rights law. In this context, the responsibilities of the actors using technology, as well as those of the technology itself, should be defined, and legal accountability mechanisms should be established in cases of violations. Furthermore, asymmetric practices that could arise if non-NATO actors use similar systems within their own ROEs could accelerate development of international norms.
Integrating ethical oversight mechanisms into technical architecture appears technically feasible. To ensure that AI operates impartially and without bias in decision-making processes, systems may need to be designed according to the principles of traceability, transparency, and accountability. Using “explainable AI” models for this purpose could ensure that decisions are open to both military and legal oversight. Thus, even scenarios where human oversight is absent, the algorithmic logic behind decisions can be retrospectively examined.
The development of dual-use sensors could necessitate common transparency protocols within NATO. The fact that these sensors, which can be used for civilian purposes on the surface, are also suitable for obtaining military intelligence could particularly increase the threat perception of non-alliance countries. Therefore, it appears feasible to develop standard agreements within NATO on matters such as sensor technology classification, data sharing, and target area limitations. Especially in hybrid warfare environments where the civil-military distinction is blurred, these protocols could play a critical role in both trust-building and crisis management.
A delicate balance will need to be struck between transparency and security in multinational sensor projects. Shared ownership of sensors, data flow schemes, and the demand for transparency regarding algorithmic decision processes can enhance trust among allies, but it can also undermine operational secrecy. To overcome this dilemma, diplomatic agreements are likely to be reached regarding not only the data collection process but also the conditions and with whom sensors will share this data. For example, technologies developed under SET-HFM-324 could be tested on platforms open only to NATO allies, achieving this delicate balance.
The deployment of dual-use sensors could necessitate new negotiations regarding host nation sovereignty. While seemingly civilian in appearance, the strategic data collection capacity of these systems could be perceived as a violation of sovereign rights by the countries where the sensors are deployed. In this context, NATO may consider establishing “sensor emplacement protocols” for future deployments, encompassing not only military but also diplomatic preliminary agreements. In particular, it seems technically and legally possible to adapt mechanisms similar to the EU’s Dual-Use Regulation practices to NATO.
The ability to collect high-resolution data could reignite discussions about data privacy. The ability of developed sensors to detect not only contamination but also human mobility, biological data, and infrastructure details could lead to new legal norms regarding the protection of personal data and the clarification of military/civilian distinctions. In this context, the “principle of data minimisation” may need to be implemented in areas where sensors are deployed. It is likely that technical guidance on this matter will be prepared within NATO and harmonised with the domestic laws of allied countries.
In conclusion, the June 2025 technology demonstration held by the NATO STO in Sweden signals a multi-layered paradigm shift detection of CBRNE threats and the monitoring of surface contamination. The use of AI-enabled sensors, in particular, and their integration with the rules of engagement have sparked discussions not only regarding technical competence but also regarding the normative framework. NATO’s advancements in this area could play a decisive role in shaping the future of multilateral security cooperation and the capacity to withstand hybrid threats.
[i] “NATO STO researchers use sensors to detect trace levels of explosives and pharmaceutical materials”, NATO Science & Technology Organization, https://shorturl.at/UjcwT, (Access Date: 02.08.2025).
