The transformation of warfare in the contemporary era is intrinsically linked to the growing centrality of information as the primary strategic asset. This qualitative shift means that military power is no longer defined solely by the number of platforms, but by the capacity to collect, process, and disseminate data in a timely manner, transforming them into knowledge that is useful for decision-making (Alberts et al., 2002).

The most notable practical expression of this new logic was the transition from Platform-Centric Warfare, whose operational effectiveness was limited to the isolated capabilities of each system, to Network-Centric Warfare (NCW). The core premise of NCW is that the electronic interconnection of geographically dispersed forces generates a synergistic effect that enhances combat power. By enabling shared situational awareness and accelerating decision speed, NCW operationalizes the principle of information superiority, seeking asymmetric advantage by integrating personnel, platforms, and formations into a single Command and Control system (Alberts et al., 2002).

However, the very conception of NCW, although revolutionary, encountered its limits as the volume of data generated by sensors, satellites, and radars grew exponentially. The natural evolution of NCW thus led to Data-Centric Warfare (DCW), a concept that deepens the previous logic by recognizing that the central contemporary challenge is not merely data sharing, but the real-time processing of massive data volumes to extract actionable intelligence (Nizienko et al., 2021).

This shift toward DCW is not exhausted by strictly operational implications. By making continuous data collection, automated processing, and algorithmic analysis central conditions of military effectiveness, DCW produces effects that extend beyond the conduct of war and directly affect how the state exercises authority over territory. It is precisely at this point that the literature situates the concept of Cyber Statecraft.

The term Cyber Statecraft describes the strategic use of digital infrastructures, data, and algorithms as core instruments of governance, security, and the projection of sovereignty in the physical world (Woods, 2017; Devanny, 2024; Stevens and Devanny, 2024). Unlike approaches that treat the digital domain as an autonomous space, this concept emphasizes its instrumental function in the exercise of state power over physical territories. From this perspective, the collection, processing, and analysis of territorial data in real time constitute a structural component of contemporary sovereignty, rather than a mere operational resource.

This dynamic does not eliminate the political dimension of sovereignty, but reconfigures it by shifting a significant portion of decision-making autonomy toward control over informational and technological infrastructures. To that extent, the capacity to mobilize information as a strategic asset becomes conditioned on the reduction of external dependencies—whether political, economic, or technical—that may impose constraints on state action (Payne, 2018; Schmidt, 2022). This issue is particularly critical for countries in the Global South, where technological dependence has historically limited decision-making autonomy in security and defense contexts.

Another issue that emerges in debates on the incorporation of artificial intelligence into the exercise of Cyber Statecraft concerns the potential generation of informational asymmetries and accountability challenges associated with the use of algorithmic systems. Studies on military AI indicate that such challenges do not arise solely from misuse, but from structural features of automation itself, such as algorithmic opacity—often described as black boxes—the embedding of biases within models, and the gradual expansion of system purposes beyond their original objectives (function creep) (Payne, 2018; Horowitz, 2018).

This combination may shift effective control over critical decisions, concentrating technical knowledge in developers or suppliers and complicating the clear attribution of responsibility within the state apparatus (Horowitz, 2018). In this context, the literature emphasizes that the consolidation of AI-based surveillance as an instrument of territorial power requires robust institutional arrangements of algorithmic governance, capable of ensuring operational transparency, continuous human oversight, and clear mechanisms of accountability (Pohle and Thiel, 2020; Horowitz, 2018).

Such safeguards are understood not as obstacles to strategic effectiveness, but as central components for preserving the legitimacy of data-mediated power, by reducing the risk of abusive or disproportionate uses of algorithmic surveillance without compromising its functionality in the fields of security and defense (Payne, 2018; Devanny, 2024).

Understanding contemporary security challenges in the Amazon requires a reappraisal of the traditional concept of security, historically centered on military defense against state-based threats. In the post–Cold War context, the international security literature incorporated non-traditional threats—of a transnational and non-state nature, such as organized crime, terrorism, and environmental degradation—which, although not manifesting as conventional conflicts, can erode state sovereignty (Buzan et al., 1998).

The transition from a social, economic, or policing problem to an issue of national defense occurs through a political process known as securitization, in which a political actor frames a given issue as an existential threat that requires emergency measures and the use of extraordinary means (Buzan et al., 1998). If the audience (public, parliament, elites) accepts this definition, the issue becomes securitized, legitimizing the use of tools normally reserved for military defense.

It is in this context that non-traditional security and securitization converge in the Brazilian case. The Amazon constitutes a region of 8.5 million km² distributed across nine South American countries, with 62% located within Brazil (Santos et al., 2024). Its low population density and dense, humid forests impose natural barriers to movement and hinder the installation of state infrastructure, creating favorable conditions for the activity of illicit actors who exploit the spatial discontinuity of state power (Instituto Brasileiro de Geografia e Estatística, 2023; Franchi, 2023). It should be noted that such challenges are not exclusive to Brazil. Colombia, Peru, and Indonesia face similar dynamics of surveillance over vast forested areas with discontinuous state presence, where these threats likewise exploit structural limitations of governance.

Not by coincidence, at least 22 criminal factions currently operate in the region, with recorded presence in 178 municipalities, equivalent to approximately 23% of the Brazilian Amazon (Fórum Brasileiro de Segurança Pública, 2023). The dynamics of these organizations reinforce the perception of an existential threat to sovereignty and territorial control, manifesting the outcome of the securitization process.

Illicit networks are structured as flexible transnational circuits, articulated around strategic points that are difficult to control—territorial nodes—often located along border zones. In these areas, illegal mining and drug trafficking operate in an integrated manner, sharing infrastructure, logistics, and trafficking routes (Fórum Brasileiro de Segurança Pública, 2023; Johnson, 2023). This capacity for spatial adaptation reveals that such networks exercise indirect forms of territorial domination, precisely by exploiting gaps in the physical presence of the state.

The configuration of this border zone, which combines thousands of kilometers of dry land, flooded areas, dense forests, and navigable rivers, hinders the imposition of effective barriers (Fórum Brasileiro de Segurança Pública, 2023). In twin cities, located on opposite sides of international borders, social, economic, and cultural ties among transboundary populations render national borders even more porous (Santos, 2016; Aguiar, 2021; Dias and Paiva, 2022; Pedreira, 2023). These borders function as zones of intense circulation of people, goods, and capital, both legal and illegal. Territorial sovereignty is not directly challenged, but progressively hollowed out by practices that escape continuous state oversight (Fórum Brasileiro de Segurança Pública, 2023).

Amazonian territorial dynamics demonstrate that governance predominantly based on permanent physical presence—through posts, patrols, and fixed infrastructure—proves inadequate in the face of a vast, mobile space exploited by highly adaptive actors. The central problem is not the absence of legal or administrative instruments, but the capacity to produce continuous territorial visibility and to efficiently activate enforcement and combat mechanisms against criminal actors and activities. In this securitized context, the state seeks tools that enable more rapid and adaptive intervention, justifying the use of means that go beyond traditional civil surveillance.

In this setting, Synthetic Aperture Radar emerges as a fundamental tool. SAR stands out for its ability to generate images of the Earth’s surface using microwaves, operating continuously and independently of weather conditions or illumination—an advantage that is decisive in regions of dense forest and persistent cloud cover (NASA Earthdata, 2025).

This capability makes it possible to generate detailed images under adverse weather conditions (rain, dense clouds, or fog), dense vegetation, and independently of sunlight—an essential attribute for the Amazon, where constant cloud cover and forest density impose significant barriers to the visual observation of illegal deforestation, clandestine airstrips, and illegal mining camps (Flores et al., 2019).

Despite its advantages, SAR presents relevant limitations for defense applications. First, the large volume of data generated requires substantial storage and processing capacity. In addition, SAR images exhibit characteristic noise known as speckle, which gives them a granular appearance and complicates visual interpretation (Meyer, 2019; Saatchi, 2019).

Although the sensor operates continuously, the extraction of useful information is not always immediate. Traditionally, this task depends on specialized analysts, who combine technical expertise with long hours of work to identify targets and patterns. In the Amazonian case, this difficulty is even greater: dense vegetation cover, irregular topography, and an extensive hydrographic network produce complex visual scenarios in which improvised airstrips, riverine vessels, or illegal mining areas can be camouflaged within the natural landscape (Ministério da Defesa, 2025a).

In light of these limitations, artificial intelligence becomes an essential instrument for enhancing the strategic value of SAR. Machine learning and deep learning algorithms automate image processing, reducing the impact of noise and identifying visual patterns that are difficult to detect through traditional human analysis (Falqueto et al., 2023). This AI–SAR integration accelerates the analytical cycle, transforming large volumes of raw data into actionable operational information in real time.

This process strengthens situational awareness, understood as an internal state of knowledge that involves the perception of relevant elements in the environment, the comprehension of their meaning in relation to the operator’s objectives, and the projection of their status into the near future (Endsley, 1995). By automating the identification of elements and dynamics within the monitored environment, AI–SAR enhances Level 1 situational awareness (Perception); by integrating and organizing these data coherently, it contributes to Level 2 (Comprehension); and by enabling inferences about trends and future behaviors, it supports Level 3 (Projection). This AI–SAR integration does not constitute the decision-making process itself, but rather functions as a critical knowledge base that underpins decision-making in complex and dynamic environments.

Beyond enhancing situational awareness, such integration can compress decision cycles associated with territorial control, particularly those described by the OODA model (Observe–Orient–Decide–Act) (Gaire, 2023). According to this model, the effectiveness of state action in complex environments depends on the ability to observe, interpret, decide, and act more rapidly and adaptively than adversarial actors (Konaev et al., 2020; Horowitz, 2018). In extensive territorial contexts marked by high uncertainty—such as the Amazon—the decision bottleneck is concentrated in the early phases of the cycle, particularly observation and orientation, which require the collection, processing, and interpretation of large volumes of spatial and temporal data.

Despite the gains associated with the compression of decision cycles, the literature on decision-support systems cautions that reducing the time between observation and action may amplify the risk of decisions based on incomplete, poorly contextualized, or excessively filtered information produced by algorithmic models (Endsley, 1995).

The distinction between data, information, and actionable intelligence thus becomes central. As Lowenthal (2017) argues, actionable intelligence is not limited to event detection, but involves contextual interpretation, relevance assessment, and integration with broader strategic objectives. AI-based systems, while efficient in identifying recurring visual patterns, operate on the basis of specific training datasets and statistical assumptions that do not capture the socioterritorial complexity of the observed space. Speed in the production of alerts does not guarantee more effective decisions and may induce premature responses when information is not adequately validated or contextualized.

Specialized analyses highlight that algorithmic errors tend to propagate more rapidly in accelerated and automated decision systems, amplifying operational and ethical risks (Morgan et al., 2020). The compression of the OODA loop through AI integration does not always produce a positive outcome. In AI–SAR territorial surveillance, decisions that rely solely on automated systems may generate a structural trade-off between speed and accuracy, in which gains in temporal efficiency are accompanied by new forms of decision vulnerability. This vulnerability constitutes an inherent risk of the securitization logic: by legitimizing the use of extraordinary and accelerated means in the name of survival, the process may inadvertently generate new dilemmas of governance and legitimacy, replacing normal policies with faster “emergency policies” that are potentially less accurate and less subject to democratic scrutiny.

The implementation of AI–SAR in Amazonian surveillance provides large-scale capabilities for continuous detection and monitoring. SAR penetrates cloud cover and operates independently of solar illumination, features that are critical for equatorial regions (NASA Earthdata, 2025). When integrated with optical remote sensing, HUMINT, and OSINT, it enables the construction of comprehensive situational awareness—the capacity to perceive, understand, and anticipate changes in the operational environment (Endsley, 1995; Couture and Toupin, 2019). In the context of territorial surveillance, this includes: (1) the identification of illicit activities (deforestation, illegal mining, invasions) in territorial nodes and twin cities; (2) the detection of behavioral patterns indicating threats; (3) the precise localization of targets; and (4) escalation forecasting based on trends (Franchi, 2023; Fórum Brasileiro de Segurança Pública, 2023). AI–SAR systems amplify these elements by automating detection, classification, and data integration.

In Brazil, CENSIPAM incorporates AI–SAR to integrate data from multiple sources, enabling continuous identification of forest changes, localization of built structures (camps, airstrips, mining sites), and tracking of movements with unprecedented precision (Franchi, 2023; CENSIPAM, 2009, 2021; Instituto Brasileiro de Geografia e Estatística, 2022). This allows for rapid responses to emerging threats, reducing windows of opportunity for illicit activities (Couture and Toupin, 2019). Similar capabilities are strategically relevant for other Amazonian countries in the Global South.

This improvement benefits both environmental protection and the territorial defense of Indigenous peoples. Early detection of invasions, deforestation, and illegal mining enables FUNAI and environmental agencies to respond before irreversible damage occurs (Franchi, 2023). Demarcated Indigenous lands present significantly lower deforestation rates, evidencing the effectiveness of Indigenous governance when supported by rapid detection and response (Virtanen et al., 2025). This integration not only enhances state surveillance, but also strengthens the self-defense of Indigenous communities by providing up-to-date information on threats (Couture and Toupin, 2019; Franchi, 2023).

However, the quality of situational awareness depends critically on training data, model robustness, and computational capacity. CNN-based models applied to Sentinel-1 SAR imagery achieve accuracy close to 90%, with AUC values above 87% when optimized (Falqueto et al., 2023; Ministério da Defesa, 2025b). Classification errors may generate false positives (incorrect identification of legitimate activities as illicit) or false negatives (failure to detect illicit activities), undermining operational effectiveness and the protection of Indigenous and local communities (Franchi, 2023).

The integration of AI–SAR not only expands surveillance, but also compresses cycles of state analysis and response. Conventional Amazonian surveillance systems rely on manual image analysis, cross-referencing with multiple databases, and interagency coordination—processes that take days or weeks. The incorporation of AI–SAR significantly reduces this interval by automating the detection of spatial and environmental anomalies, target and structure classification, within much shorter timeframes, often measured in hours (Ministério da Defesa, 2025b; Centro Gestor e Operacional do Sistema de Proteção da Amazônia, 2009, 2021). This temporal reduction transforms the dynamics between surveillance and action.

This action–reaction dynamic is embedded in the strategic transformation known as data-centric warfare (Nizienko et al., 2021). Military concepts such as the OODA loop emphasize that strategic advantage belongs to those who process information and decide more rapidly than their adversaries (Johnson, 2023; Gaire, 2023). AI–SAR optimizes each stage: continuous and automated observation; accelerated orientation through predictive analysis; decisions informed by algorithmic recommendations; and coordinated action through integrated systems. The result is a temporal reduction of the decision cycle, altering the speed at which the state responds to threats (Centro Gestor e Operacional do Sistema de Proteção da Amazônia, 2009, 2021).

This compression offers operational advantages. Rapid responses to illegal deforestation reduce environmental damage; efficient interdiction of drug and arms trafficking strengthens public and international security; accelerated detection of illegal mining protects natural resources and the rights of Indigenous peoples. However, this acceleration generates a structural trade-off between speed and precision (Endsley, 1995; Lowenthal, 2017), as temporal gains amplify the risks of algorithmic error (Horowitz, 2018; Payne, 2018; Morgan et al., 2020). Decisions based solely on algorithmic recommendations reduce the time available for human review, contextualization, or correction (Furtado et al., 2024).

Algorithms may misclassify activities—identifying legitimate actions as suspicious, or vice versa. The literature shows that algorithmic errors propagate more rapidly in accelerated decision-making contexts, amplifying operational risks when adequate verification mechanisms are absent (Morgan et al., 2020; Endsley, 1995). An authorized controlled burn for forest management, if erroneously flagged as illegal deforestation, could trigger preliminary administrative responses. Dual-verification protocols—manual analysis, consultation with local communities, and prior review—can mitigate these risks, ensuring that state interventions are grounded in an appropriate assessment of the legitimacy of the activity (Centro Gestor e Operacional do Sistema de Proteção da Amazônia, 2009, 2021).

The integration of AI–SAR optimizes human and financial resources by reducing operational costs and the need for permanent personnel mobilization in territorial monitoring. These gains may strengthen environmental protection and rationalize public spending. However, their impact is not automatic and depends on the institutional arrangements that regulate decision cycles.

In contexts where such compression occurs without safeguards for transparency, auditability, and control, technological efficiency may generate institutional asymmetries, shifting the balance between state capacity and democratic oversight. Thus, the quality of institutional use becomes as relevant as technical capabilities themselves.

The development of Lessonia-1 represents a significant achievement in Brazilian technological independence (Ministério da Defesa, 2022; Couture and Toupin, 2019; Galdino, 2022). Historically, Brazil has depended on foreign suppliers for critical surveillance technologies, creating geopolitical vulnerability through access denial, the imposition of political conditions, or mandatory data sharing. Indigenous capabilities in remote sensing and AI algorithms provide autonomy vis-à-vis external suppliers, allowing control over critical surveillance infrastructure.

Such independence is strategically significant in the context of geopolitical competition for technological supremacy. Powers such as the United States and China invest heavily in AI and remote sensing as strategic assets (Couture and Toupin, 2019; Schmidt, 2022). For Brazil, developing AI–SAR capabilities reduces external dependence and affirms technological sovereignty (Azevedo et al., 2021), particularly in the Amazon, where foreign geopolitical interests—environmental, economic, and strategic—frequently conflict with national interests.

However, technological autonomy does not automatically guarantee the implementation of AI–SAR aligned with democratic and inclusive values (Pohle and Thiel, 2020). Brazil may manage SAR image acquisition and algorithm development, but the quality of this implementation depends on regulatory frameworks that define how the technology is applied and who participates in decision-making processes. The inclusion of Indigenous peoples and Amazonian communities in governance processes strengthens the legitimacy of public policies and incorporates territorial knowledge essential for more effective and context-sensitive monitoring.

The autonomous operation of Lessonia-1 initiated in 2025 (Força Aérea Brasileira, 2025; Centro Gestor e Operacional do Sistema de Proteção da Amazônia, 2009, 2021; Virtanen et al., 2025) allows Brazil full control over the collection and processing of SAR imagery, positioning the country as a regional reference for Global South states seeking to reduce external dependence in territorial surveillance.

SAR imagery reveals strategic infrastructure and patterns of territorial occupation that require robust protection against unauthorized access. Such protection constitutes an institutional and political challenge, not merely a technical one: it requires regulation governing data access, safeguards against commercialization, and audit mechanisms that ensure transparency. Thus, full technological sovereignty requires not only independence in technological development but also robust data governance capable of protecting national security and the territorial rights of Indigenous peoples.

The implementation of AI–SAR in Amazonian surveillance requires robust algorithmic governance—processes, norms, and institutions that regulate the development, deployment, and oversight of AI systems, ensuring that automated decisions remain aligned with democratic values and fundamental rights (Pohle and Thiel, 2020). It is particularly important to consider how the securitization of the Amazon may influence how such systems are applied, requiring safeguards against misuse and function creep (Buzan et al., 1998; Brayne, 2020; Jobin et al., 2019).

Effective governance rests on four pillars: (1) operational transparency—explaining how the system identifies targets and generates alerts; (2) accountability—clear responsibility for errors and inappropriate decisions; (3) mitigation of algorithmic bias—ensuring that models do not induce discriminatory processes; and (4) protection of human rights and privacy—mechanisms against selective surveillance and data misuse (Furtado et al., 2024; Islam and Wasi, 2024; Kazmi, 2024).

In Brazil, the coordination of Amazonian territorial surveillance through remote sensing falls under the responsibility of CENSIPAM (Centro Gestor e Operacional do Sistema de Proteção da Amazônia, 2009, 2021). which integrates data from multiple sources, including satellites such as Lessonia-1, to monitor threats to territorial integrity. Although subject to audit and oversight mechanisms, these are not sufficient to ensure comprehensive ethical governance, particularly when operations may be affected by erroneous classifications (Franchi, 2023).

In this regard, the procedural rights of Indigenous communities—particularly the right to participate in the review of algorithmic decisions, to present evidence of the legitimacy of their practices, and to receive reasoned decisions—are safeguarded by institutions such as FUNAI and non-governmental organizations working to defend territorial and cultural rights (Jobin et al., 2019). The effectiveness of these institutions depends on infrastructure—remote sensing, monitoring equipment, and data-processing capabilities—that enables operations at a territorial scale (FUNAI, 2025; Virtanen et al., 2025; Furtado et al., 2024). Brazil has the largest Indigenous population in Latin America and the largest extent of demarcated Indigenous lands (Virtanen et al., 2025).

Beyond this, AI–SAR governance also involves protecting sensitive data against unauthorized access, commercialization, or misuse, requiring regulation on who may access data, under what conditions, and with which safeguards (Couture and Toupin, 2019; Pohle and Thiel, 2020; Falqueto et al., 2023). Without robust technological infrastructure to sustain these frameworks, independence in remote sensing (Lessonia-1) and operational coordination (CENSIPAM) would not guarantee adequate protection of Indigenous communities’ procedural rights: Brazil would possess technical sovereignty in image collection, but might lack capabilities in processing, storage, and computational systems that enable ethical controls, transparent auditing, and contestation (Couture and Toupin, 2019).

The central question is whether Brazil will achieve full technological autonomy—not only in remote sensing and algorithms, but also in critical infrastructure for processing, storage, and computational systems—and whether this independence will be accompanied by ethical governance frameworks that ensure use consistent with human rights and the protection of local communities (Virtanen et al., 2025).