1. Introduction

Outer space is an inherently hazardous environment. Extreme temperatures, radiation exposure, orbital debris, and the immense speed at which objects travel mean that even minor technical failures may lead to catastrophic consequences. The margin for error is exceptionally narrow, and incidents rarely remain confined to a single actor. Damage caused by one satellite may cascade across orbital regimes, affecting multiple operators and potentially even activities on Earth^[1]. In such an environment, the structural risks of space significantly heighten the regulatory importance of prevention and oversight.

At the same time, the legal governance of outer space is being tested by profound technological and commercial change. Rapid innovation, declining launch costs, and the rise of private operators have transformed the space sector from a predominantly state-led domain into a far more complex ecosystem of public and private activity^[2]. Private mega-constellations, in-orbit servicing missions, robotic satellite platforms, and space resource initiatives are no longer speculative concepts, but operational realities. This transformation has increased both the scale of space activity and the difficulty of supervising it effectively^[3].

A further source of regulatory complexity is the growing integration of artificial intelligence into space systems. AI technologies are increasingly used for autonomous navigation, collision avoidance, mission optimization, data processing, and other operational functions. While these systems offer clear benefits in terms of efficiency, resilience, and responsiveness, they also introduce new forms of opacity, complexity, and unpredictability. As operational control is delegated to adaptive or partially autonomous systems, traditional assumptions about human oversight, traceability, and foreseeability of harm become more difficult to sustain.

These developments acquire particular significance in light of broader normative concerns surrounding artificial intelligence. Across legal and policy debates, questions of accountability, responsibility, and transparency have emerged as central challenges of AI governance. When decision-making processes are partially autonomous and technically non-transparent, it becomes increasingly difficult to determine who should be answerable for harmful consequences, how responsibility should be allocated, and whether algorithmic behaviour can be meaningfully scrutinized or contested. In the context of outer space – where harmful consequences may be transboundary, irreversible, and system-wide – these concerns become especially acute^[4][5].

Taken together, these developments converge on a common regulatory problem. The orbital environment demands robust ex ante governance, yet the systems increasingly deployed within it are autonomous, complex, and difficult to supervise through traditional means. Unlike many terrestrial domains, harm in outer space is not easily contained or remedied. A single malfunctioning satellite, one defective autonomous manoeuvre, or one compromised AI-enabled object may generate debris or disruption that affects all operators, regardless of nationality. In such circumstances, regulatory fragmentation itself becomes a source of systemic risk^[6].

This, in turn, raises a broader legal question: whether the existing framework of international space law is capable of responding to these emerging governance challenges. International space law is often criticized as outdated and insufficiently responsive to contemporary commercialization and technological transformation^[7]. The core treaties were drafted in an era dominated by state-led space activity, long before the rise of private mega-constellations and AI-enabled autonomous systems^[8]. That critique is not unfounded. Yet it may overlook an important structural feature of the existing legal architecture: the allocation of responsibility and liability to states for national activities in outer space.

This article argues that, despite widespread criticism of international space law as outdated, its state-centric allocation of responsibility and liability may still offer a meaningful structural foundation for the ex ante governance of artificial intelligence in outer space. In particular, Article VI of the Outer Space Treaty requires states to authorize and continuously supervise the activities of non-governmental entities, thereby embedding ex ante control and ongoing oversight into the legal architecture of space governance. Complementing this framework, Article VII attributes international liability for damage caused by space objects to the launching state. Together, these provisions create a dual mechanism of responsibility and liability: states must not only supervise private actors, but also internalize the legal and financial consequences of potential harm.

Far from being merely an obsolete feature of an earlier era, this state-centric design may offer certain regulatory advantages in the context of AI-enabled space activities. Because access to orbit depends on identifiable launch infrastructure, licensing procedures, and national authorization mechanisms, the space sector remains more institutionally mediated than many terrestrial domains in which AI is deployed. Moreover, the catastrophic and potentially cascading consequences of orbital incidents create strong incentives for preventive regulation. The combination of physical bottlenecks, systemic fragility, and treaty-based state responsibility may therefore make international space law more adaptable to AI-related risks than is often assumed.

This article therefore examines whether the existing international legal framework governing outer space is capable of addressing the risks associated with artificial intelligence and autonomous systems as applied within the space demain/industry. It argues that, although the current regime was not designed with AI in mind, its emphasis on authorization, continuing supervision, and state responsibility provides an important basis for ex ante governance. Rather than treating international space law solely as an outdated framework in need of replacement, the article suggests that it also contains underappreciated structural resources for regulating emerging technologies in an exceptionally hazardous environment.

2. Foundations of International Space Law

International space law is founded on five key multilateral treaties adopted under the auspices of the United Nations through the Committee on the Peaceful Uses of Outer Space (COPUOS). Concluded between 1967 and 1984, these treaties established the basic legal architecture that continues to shape both international space governance and national space legislation. Together, they articulate fundamental principles governing the exploration and use of outer space, allocate rights and obligations among states, and provide core rules on matters such as responsibility, liability, rescue, and registration^[9].

The five core treaties are the Treaty on Principles Governing the Activities of States in the Exploration and Use of Outer Space, including the Moon and Other Celestial Bodies (1967, “Outer Space Treaty”), the Agreement on the Rescue of Astronauts, the Return of Astronauts, and the Return of Objects Launched into Outer Space (1968, “Rescue Agreement”), the Convention on International Liability for Damage Caused by Space Objects (1971, “Liability Convention”), the Convention on Registration of Objects Launched into Outer Space (1976, “Registration Convention”), and the Agreement Governing the Activities of States on the Moon and Other Celestial Bodies (1984, “Moon Agreement”). While all five treaties together form the historical and normative basis of international space law, the first four continue to carry the greatest practical significance in contemporary space governance.

At the same time, the foundational treaties were drafted in a very different technological and institutional context. They were designed primarily for a state-centric and largely non-commercial era of space activity, long before the rise of private mega-constellations, rapid innovation cycles, and AI-enabled autonomous systems. For this reason, scholars and practitioners have often criticized the treaty framework as insufficiently responsive to contemporary regulatory challenges, particularly in relation to private actors and complex liability scenarios^[10].

For the purposes of this article, particular attention must be paid to the distinction between state responsibility and state liability within the legal framework established by the Outer Space Treaty and the Liability Convention. Although these concepts are frequently translated into some languages by a single term, they represent distinct legal categories serving different functions within international law^[11]. Article VI of the Outer Space Treaty is based on the concept of responsibility and requires States to ensure that national activities in outer space, including those conducted by non-governmental entities, comply with international law. In practice, this obligation is implemented through the requirement to authorize and continuously supervise private actors. By contrast, Article VII adopts the concept of liability for damage caused by space objects and establishes that the launching State bears international liability for damage caused by such objects irrespective of whether the relevant activity was carried out by governmental or private actors^[12].

Questions of international responsibility in general, and of responsibility and liability in space law in particular, have been the subject of considerable scholarly debate, and any analysis of Article VI of the Outer Space Treaty must engage with that literature rather than treat the provision’s meaning as settled. The prevailing view construes Article VI as imposing a due diligence obligation on States: States are required to adopt reasonable measures to authorise, monitor, prevent and, where appropriate, sanction unlawful conduct by private actors, without incurring automatic responsibility for each such act as such^[13]. This interpretation is consistent with the general law of state responsibility, which does not ordinarily attribute the conduct of private actors to the State but instead imposes obligations relating to prevention, supervision and control. A broader – and contested – interpretation, advanced by Bin Cheng, suggests that the conduct of private entities engaged in outer space activities may be regarded as attributable to the State, such that a breach of international law by a non-governmental operator may engage state responsibility in a manner approaching direct state conduct^[14].

The doctrinal uncertainty surrounding the interpretation of Articles VI and VII of the Outer Space Treaty, particularly regarding the scope of state obligations and the attribution of responsibility and liability for resulting harm, presents difficulties even in the context of conventional space activities. Divergent interpretations concerning the extent of supervisory duties under Article VI and the operation of liability mechanisms under Article VII already generate uncertainty in relation to traditional space activities involving private actors and a substantial degree of human decision-making and control. These challenges become even more pronounced in the context of artificial intelligence. The deployment of autonomous and adaptive systems in outer space may complicate the assessment of conduct, causation and oversight, particularly where the operation of such systems involves limited, indirect or evolving forms of human intervention. Existing legal uncertainty concerning the attribution of conduct and responsibility may therefore be further amplified where AI systems generate outcomes that are difficult to predict, explain or control ex ante.

The purpose of this article is not to resolve the broader doctrinal debate concerning the interpretation of Articles VI and VII of the Outer Space Treaty or to advocate for a particular understanding of state responsibility and liability in space law. Rather, this article proceeds from the premise that the existence of competing interpretations itself constitutes a relevant legal risk that must be taken into account when addressing the governance of AI systems in outer space. Irrespective of which interpretation ultimately prevails, the absence of a settled legal understanding may significantly affect regulatory design, supervisory expectations and the allocation of legal responsibility for harm caused by AI-enabled space activities.

In principle, it would be best if the treaty regime could be modernized or supplemented by new binding international agreements better tailored to contemporary space activities. In practice, however, the negotiation of new multilateral treaties in this field has proven politically difficult^[15]. As a result, much recent normative development has taken place through non-binding instruments, guidelines, and other forms of soft law adopted within the UN framework, including the COPUOS Space Debris Mitigation Guidelines. Although such instruments do not carry the force of law, they play an important role in shaping expectations of responsible conduct, particularly in areas such as orbital safety and debris mitigation. At the same time, the regulatory center of gravity has increasingly shifted toward national legislation, as states develop domestic licensing and supervisory frameworks for commercial space activities. This trend reflects a pragmatic response to the limits of treaty reform and highlights the continuing importance of state-based governance in the contemporary regulation of space activities^[16].

3. From Obsolete to Essential: Leveraging International Space Law for AI Governance

Before examining how the structural features of space governance may be leveraged for AI oversight, it is necessary to clarify the relationship between the international and domestic legal levels. International space law – most importantly the Outer Space Treaty and the Liability Convention – does not itself establish detailed rules. Rather, it creates a main framework within which states regulate national space activities through domestic law (i.e. allocation of responsibility and liability)^[17]. It is therefore at the domestic level that authorization procedures are designed, technical and safety standards are specified, supervisory mechanisms are created, and conditions for the deployment of autonomous systems are defined. This distinction is especially important in the AI context, because autonomous systems challenge traditional assumptions about foreseeability, controllability, and human oversight – assumptions that are relevant both to the international allocation of responsibility and liability and to the domestic operationalization of authorization and supervision. This dual-level structure is sometimes underappreciated in critiques of international space law.

A common objection in the literature is that the space treaties, having been drafted in the second half of the twentieth century, are poorly suited to contemporary commercialization and AI-driven operations. That criticism is partly justified: the treaties were not designed with machine learning, autonomous decision-making, or private mega-constellations in mind. Yet such critiques may overstate the consequences of that mismatch if they assume that the international framework must itself contain the full substantive content of AI regulation. Its primary function is instead structural: to allocate responsibility, impose supervision obligations, and create incentives for states to build domestic regulatory systems. In this respect, the question is not simply whether international space law contains AI-specific rules, but whether its institutional design is capable of supporting AI governance through domestic implementation. It is this latter question that is examined here.

The argument advanced in this section is that the state-centric architecture of international space law may provide more regulatory value than is often acknowledged. This is not because the treaties offer a complete or technologically tailored response to AI. They clearly do not. Rather, their continuing significance lies in the way they channel regulatory authority through states and link that authority to legal and financial consequences. In particular, Article VI of the Outer Space Treaty obliges states to authorize and continuously supervise the activities of non-governmental entities, while the liability regime gives states strong incentives to prevent harm before it occurs. For the purposes of AI governance, this structure matters because the deployment of autonomous systems in orbit remains dependent on highly concentrated physical infrastructure and on national authorization processes capable, at least in principle, of imposing ex ante conditions. The discussion below develops this claim in three steps: first, it identifies the structural bottleneck created by orbital access; second, it explains how this bottleneck enables ex ante oversight; and third, it considers how the fragility of the orbital environment reinforces incentives for strict supervision.

3.1 Orbital Access as a Structural Bottleneck

Although technological progress has significantly reduced some of the costs associated with space activities, especially in satellite manufacturing and standardized components, access to orbit remains subject to substantial technological, financial, and infrastructural barriers. The relevant comparison with terrestrial AI is therefore instructive. Most terrestrial AI systems can be developed, modified, and deployed with little or no governmental awareness, and often without passing through any centralized institutional checkpoint. By contrast, an AI-enabled space object cannot enter orbit without reliance on launch services, launch infrastructure, licensing, and state authorization. In this respect, outer space remains a highly mediated environment rather than an open-access domain.

Launch capabilities continue to be concentrated among a relatively limited number of actors worldwide, and the number of geographically suitable and operational launch sites is inherently constrained. Effective orbital launches require particular geographic, safety, and logistical conditions, which not every territory can provide. This concentration has direct regulatory significance. Any operator seeking to place an AI-enabled satellite into orbit must depend on a launch vehicle and launch site functioning under some form of state authority or state-approved regulatory framework. Moreover, under international law, the concept of the “launching state” is interpreted broadly enough to extend legal accountability beyond the single state from whose territory a launch physically takes place. States that procure, authorize, or conduct launches may all become relevant actors within the responsibility and liability framework.

This concentration of access creates a structural bottleneck, and that bottleneck may be normatively valuable. Much of the literature presents the state-centric structure of space law as a residue of an earlier era, when space activity was dominated by governments rather than private actors. Yet in the context of AI governance, that same structure may serve as a regulatory advantage. Because access to orbit is not freely scalable, states retain a meaningful opportunity to intervene before deployment occurs. In practical terms, autonomous systems cannot simply be “released” into orbit in the same manner as terrestrial software can be deployed online. They must pass through a state-regulated gateway.

3.2 Ex Ante Oversight Through Domestic Authorization

The existence of this bottleneck does not by itself guarantee effective governance. Its importance lies in the regulatory opportunities it creates at the domestic level. Because states remain the legal focal point of responsibility and liability under international law, they have both the authority and the incentive to condition launch approvals on compliance with specific authorization criteria. Those criteria may include safety testing, fail-safe requirements, cybersecurity protections, software verification, audit documentation, or disclosure obligations tailored to AI-enabled systems. The concrete design of such measures is a matter of domestic law, but the incentive to create them is reinforced by the international framework.

This is where the interaction between international and domestic law becomes most important. International law does not prescribe the full content of AI governance, but it does shape the behavior of states by exposing them to reputational, legal, and potentially financial consequences if inadequately supervised national activities cause harm. To that extent, the treaties may be understood not as self-sufficient regulatory codes, but as framework instruments that externalize pressure onto domestic authorization systems. This is especially significant in the AI context, where ex post responses may be insufficient. Once an autonomous system is deployed in orbit, errors may be difficult to predict, difficult to interrupt, and impossible to reverse in practical terms. For high-risk orbital activities, the preventive dimension of governance therefore becomes far more important than in many terrestrial settings.

A possible objection is that this argument assumes too much regulatory capacity on the part of states. Not all states possess equivalent levels of technical expertise, administrative resources, or institutional experience in assessing AI-enabled systems. That concern is well founded and should not be minimized. However, it does not negate the structural point being made here. The claim is not that all states currently exercise robust and effective supervision, but that the architecture of space law creates identifiable control points through which such supervision can in principle be organized. In other words, the framework may be institutionally imperfect and unevenly implemented, yet still structurally better positioned for ex ante AI oversight than many terrestrial legal settings in which no comparable bottleneck exists.

This point also bears on the scholarly debate over whether the current regime is functionally obsolete. Critics are right to observe that the treaties do not directly regulate autonomous decision-making, machine learning models, or AI validation procedures. But it does not necessarily follow that they are irrelevant. If the central problem of AI governance in space is how to ensure that high-risk systems are scrutinized before deployment, then a legal framework that channels deployment through state authorization and continuing supervision may retain significant regulatory value even without AI-specific treaty language. The present argument therefore departs from stronger obsolescence narratives by suggesting that the existing system may still provide an adaptable institutional foundation for domestic AI governance.

3.3 Orbital Fragility and the Incentive for Strict Supervision

The plausibility of strong ex ante oversight is further reinforced by the physical characteristics of the orbital environment itself. Outer space is not merely dangerous in the ordinary sense; it is structurally fragile. Harm is cumulative, difficult to contain, and potentially irreversible. The phenomenon commonly described as the Kessler syndrome illustrates this logic particularly clearly: a single collision or technical malfunction may generate debris that triggers cascading collisions, degrading or even rendering unusable particular orbital regimes. In such a context, the risks associated with AI-enabled systems are not limited to isolated operator error. Autonomous decision-making, software faults, cybersecurity failures, or poorly supervised updates may all contribute to events with systemic and transboundary consequences.

This feature of orbital activity strengthens the case for demanding standards of authorization and supervision. In legal terms, it supports the view that even if state responsibility under Article VI is interpreted through a due diligence lens, the content of that due diligence may have to be especially exacting in the context of AI-enabled space activities. Put differently, the significance of the responsibility/liability debate is not merely conceptual. It bears directly on the level of oversight that should be expected where the consequences of failure may affect the entire orbital commons. In this respect, the orbital environment itself provides an important reason to read the obligations of authorization and supervision as substantively robust rather than purely formal.

Comparable patterns can be observed in other highly regulated sectors characterized by low-probability but high-impact failures. Nuclear energy and systemic financial regulation offer familiar examples. In both contexts, isolated technical or institutional failures have produced cascading, transboundary, and long-term harm, leading states to adopt dense frameworks of licensing, preventive oversight, and continuous monitoring. These sectors are often criticized as heavily regulated, yet such regulatory intensity is widely understood as a proportionate response to systemic risk. Although orbital incidents are less frequent than nuclear accidents or financial collapses, their potentially irreversible consequences justify a similarly preventive orientation.

None of this means that the state-centric model is free from difficulty. It depends on administrative expertise, international coordination, and the willingness of states to enforce demanding authorization standards even where commercial pressures point in the opposite direction. Nor does it eliminate the challenge that AI systems may evolve post-launch or be updated remotely in ways not fully captured by initial review. Nevertheless, these limitations do not undermine the central point. Far from being merely obsolete, the state-centered architecture of international space law may provide a concentrated and practically meaningful framework for the ex ante governance of AI in orbit. Its value lies not in offering a complete substantive code for artificial intelligence, but in creating legal incentives, institutional checkpoints, and regulatory leverage at the points where access to orbit can still be controlled.

While this model is not without institutional challenges, it demonstrates that the state-centric structure of space law does not necessarily represent regulatory obsolescence. On the contrary, it provides a concentrated regulatory checkpoint that, if effectively utilized, may function as a pragmatic mechanism for governing autonomous systems in outer space.

An additional structural factor reinforcing the plausibility of strong state oversight is the extreme fragility of the orbital environment. Unlike most terrestrial domains, outer space is characterized by cumulative and potentially irreversible externalities^[18]. The phenomenon commonly referred to as the Kessler syndrome illustrates how even a single collision or technical malfunction may trigger a cascading chain reaction of debris generation, ultimately rendering certain orbital regimes unusable^[19]. Given the autonomous decision-making of AI systems, errors may trigger or accelerate debris generation predicted by the Kessler model. In such a scenario, the consequences are not confined to a single operator or state; rather, they affect the entire international community. Access to orbit is therefore structurally interdependent: either it remains viable for all actors, or it may become effectively inaccessible to all. This “all-or-nothing” characteristic significantly increases the systemic risk associated with autonomous systems operating at scale.

Comparable risk structures can be observed in other highly regulated sectors where small failures may produce disproportionate systemic consequences. Nuclear energy provides a paradigmatic example: the accidents at Chernobyl Nuclear Power Plant and Fukushima Daiichi Nuclear Power Plant demonstrated that localized technical failures can have transboundary and long-term effects^[20]. Similarly, in the financial sector, the collapse of Lehman Brothers during the 2008 crisis revealed how interconnected systems amplify localized risk into global instability^[21]. In both sectors, states responded by establishing comprehensive frameworks of ex ante supervision, licensing, and continuous monitoring. These industries are frequently described as “overregulated,” yet such regulatory intensity is widely recognized as necessary and proportionate to the magnitude of potential harm. While incidents in orbital space occur less frequently than nuclear or financial crises, their systemic and potentially irreversible consequences justify a similarly intensive regulatory approach.

4. Challenges: Assessing the Limits of the Proposed Framework

The preceding analysis has argued that the state-centric architecture of international space law, reinforced by the physical bottlenecks of orbital access, may provide a meaningful foundation for the ex ante governance of AI-enabled space systems. That argument, however, should not be overstated. A governance model may appear coherent at the structural level yet still encounter serious difficulties in practice. It is therefore necessary to assess not only the strengths of the proposed framework, but also its institutional limits, potential points of erosion, and broader policy trade-offs^[22].

This section addresses three principal challenges. The first concerns whether states possess the technical expertise and administrative capacity necessary to evaluate increasingly complex AI systems within authorization regimes originally designed for more conventional space technologies. The second asks whether the proposed model may weaken over time if the number of orbital actors continues to expand and commercialization further accelerates. The third concerns the danger that a precautionary, authorization-based approach may impede innovation by imposing excessive regulatory burdens on emerging technologies. Each of these objections deserves careful consideration. None is fatal to the proposed framework, but each reveals important tensions that any realistic model of AI governance in outer space must confront.

4.1 Expertise Gaps and Institutional Capacity

A central institutional weakness of the proposed ex ante model lies in the limited technical expertise available within many national administrations. AI systems – particularly those deployed in safety-critical orbital environments – are complex, adaptive, and often opaque. Their meaningful assessment may require interdisciplinary competence in machine learning, software assurance, cybersecurity, orbital mechanics, and systems reliability. Yet many national authorities responsible for space authorization were not originally designed to evaluate high-dimensional AI models or autonomous decision architectures^[23]. They were created primarily to assess more conventional questions of launch safety, registration, licensing, and technical compliance.

The resulting risk is that the existence of an authorization framework may create only the appearance of regulatory control. Without sufficient internal expertise, or without access to independent technical review, states may lack the capacity to scrutinize AI-enabled systems in a substantively meaningful way prior to launch. In that scenario, authorization becomes formal rather than effective. This is a serious objection, and it underscores that the viability of ex ante governance depends not only on legal structure, but also on institutional capability.

At the same time, the weakness should not be assessed in isolation from the incentive structure created by international space law. Under the liability regime applicable to space activities, launching states may face significant legal and financial consequences if inadequately supervised space objects cause damage. Although liability rules differ depending on the location and nature of the damage, the broader point remains: orbital accidents, debris-generating collisions, and loss of high-value infrastructure may produce exceptionally costly consequences^[24]. From this perspective, investment in ex ante expertise may be economically rational. Even costly mechanisms such as expert review panels, third-party certification, technical audits, or specialized AI assessment units are likely to be less expensive than the consequences of catastrophic orbital harm.

A further challenge arises from disparities among states. Not all launching states possess comparable scientific infrastructure, administrative capacity, or access to expert communities. This asymmetry may produce uneven supervisory standards and create weak links in the overall system^[25]. Here, too, the objection is real. The state-centric model does not ensure uniform quality of oversight across jurisdictions. Yet the shared and interdependent nature of the orbital environment also creates strong incentives for cooperation. Because the costs of a poorly supervised launch are not confined to the authorizing state, orbital safety functions as a collective good. This suggests that technical expertise, model standards, and supervisory practices may need to be shared across jurisdictions if ex ante governance is to remain credible in practice. The challenge of expertise therefore qualifies the proposed model, but does not necessarily defeat it; rather, it indicates that effective state-based governance may require institutional support mechanisms beyond the state acting alone^[26].

4.2 Growth in Orbital Actors and the Durability of Ex Ante Control

A second objection is more structural. Even if state-based authorization currently provides a workable checkpoint, this may not remain true as access to space becomes cheaper and the number of private operators grows. If orbital activities continue to expand rapidly, states may find it increasingly difficult to exercise meaningful oversight over all AI-enabled systems deployed by their national operators. Regulatory capacity is finite, while the complexity of machine-learning-based systems may place increasing pressure on already limited administrative resources. On this view, the bottleneck that currently supports ex ante control may gradually weaken as commercialization accelerates.

This concern deserves serious attention, but its empirical basis should not be overstated. Some parts of the space sector – most notably small-satellite manufacturing and standardized components – have indeed become more affordable^[27]. Yet launch capability has not undergone a comparable degree of democratization. The development, certification, and operation of orbital launch vehicles remain technologically sophisticated, capital-intensive, and heavily regulated. Access to orbit therefore continues to depend on a relatively small number of launch providers and operational launch sites. Even if the number of payload developers and satellite operators increases substantially, this does not necessarily imply a corresponding fragmentation of the launch segment^[28].

Indeed, market dynamics may point in the opposite direction. In infrastructure-heavy sectors, declining costs in one segment often produce concentration rather than unlimited entry in another. Launch services may increasingly resemble a centralized transport function, in which a limited number of providers offer capacity to a growing number of users. In this respect, commercial developments such as “rideshare” or “launch-as-a-service” models may actually reinforce, rather than undermine, the structural concentration through which regulation can operate. Operators investing substantial sums in AI-enabled payloads are also likely to prefer established launch providers with demonstrated reliability, rather than untested entrants. This preference further supports concentration in the launch market^[29].

The significance of this point is regulatory rather than merely economic. Even if the number of actors designing satellites increases, access to orbit may remain funneled through a relatively small and identifiable set of gateways. If so, the state does not need to supervise every possible technological actor in the abstract; it needs to ensure that no payload reaches orbit without passing through authorization and launch control processes. This creates a structural asymmetry that distinguishes the space context from most terrestrial AI applications, where systems may be developed and deployed with minimal governmental visibility.

That said, the objection cannot be dismissed entirely. If launch markets were to become substantially more decentralized, or if new forms of access to orbit were to erode the present concentration of infrastructure, the effectiveness of the bottleneck model could diminish over time. The claim advanced here is therefore contingent rather than absolute. It is not that the bottleneck will necessarily remain intact indefinitely, but that under current technological and market conditions it remains sufficiently concentrated to support meaningful ex ante governance.

Where such concentration persists, domestic authorization frameworks can operate as practical tools of control. States may require payload operators to obtain authorization before seeking launch services, to demonstrate compliance with safety, cybersecurity, and technical reliability requirements, and, where necessary, to submit documentation, testing results, or audit materials relevant to AI-enabled functions. Launch providers, in turn, may be required to verify the existence of valid authorization before carrying payloads to orbit. In this way, the authorization process can function not merely as a formal licensing step, but as an upstream checkpoint through which deployment is conditioned on prior scrutiny.

Ex ante review, however, does not eliminate all risk. AI systems may evolve post-launch, may be remotely updated, or may behave in unforeseen ways despite pre-deployment assessment. For that reason, authorization should not be understood as a complete solution. Its importance lies rather in reducing the probability that inadequately assessed high-risk systems reach orbit in the first place, while also creating the legal and procedural basis for continuing supervision after deployment. In this respect, upstream control remains valuable even where downstream uncertainty cannot be fully removed.

4.3 Innovation Costs and the Risk of Regulatory Overreach

A third challenge concerns the costs of precaution itself. An authorization-based governance model may reduce the likelihood of harmful deployment, but it may also slow the pace of innovation. Space activities have developed rapidly in recent decades, particularly as private actors have entered the market and technical development cycles have accelerated. More demanding authorization procedures for AI-enabled systems could lengthen deployment timelines, increase compliance costs, and reduce flexibility for operators. These effects may be especially significant for smaller or newer market entrants^[30].

A related concern is that states, motivated by the prospect of liability and by the catastrophic potential of orbital accidents, may adopt overly conservative authorization criteria. If regulatory institutions become excessively risk-averse, they may block or delay the deployment of technologies whose risks are uncertain but manageable. In this sense, the very features that make ex ante control attractive from a safety perspective may also produce a chilling effect on innovation. This is a genuine policy tension and should not be underestimated^[31].

At the same time, the existence of such trade-offs does not mean that precautionary governance is misguided. In sectors characterized by high-impact, potentially irreversible harms, slower deployment may sometimes be a justified cost of responsible regulation. The relevant question is not whether oversight imposes burdens – it inevitably does – but whether those burdens are proportionate to the magnitude of the risks at stake. In the orbital environment, where a single failure may generate long-term debris or compromise access for multiple actors, some degree of regulatory friction may be not only acceptable, but necessary.

The challenge, then, is one of calibration rather than rejection. If authorization mechanisms are to remain both legitimate and effective, they should be designed to distinguish between unacceptable risk and manageable uncertainty. This may require proportionate review procedures, staged approvals, experimental licensing, expert advisory bodies, and mechanisms for iterative testing under controlled conditions. Such approaches would allow states to preserve the preventive logic of ex ante governance while reducing the danger that authorization becomes an instrument of excessive conservatism^[32].

The broader implication is that the proposed framework should not be understood as advocating rigid or static control. Its value lies in establishing legal and institutional checkpoints through which high-risk autonomous systems can be evaluated before deployment, while still leaving room for differentiated and adaptive regulatory techniques. The challenge of innovation, therefore, does not refute the ex ante model. Rather, it highlights the need to design it carefully.

4.4 Interim Assessment

Taken together, these three challenges show that the proposed framework is neither automatic nor cost-free. Its effectiveness depends on institutional expertise, continued concentration in access to orbit, and the careful calibration of precautionary oversight. These are meaningful constraints, and they should temper any overly optimistic account of state-centered AI governance in outer space. Yet they do not alter the central conclusion of this article. Even acknowledging its limits, the state-centric structure of international space law remains more adaptable to ex ante AI governance than critiques of obsolescence often suggest. Its value lies not in eliminating uncertainty, but in providing legal incentives, regulatory leverage, and institutional control points through which uncertainty can be managed before harm occurs.

5. Conclusion

In sum, the regulatory landscape of outer space reflects a complex interplay between extraordinary environmental hazards, rapid technological advancement, and evolving legal frameworks. The combination of high systemic risk, AI integration, and increasing private sector participation underscores the urgent need for proactive governance measures. Ex ante authorization, embedded within the state-centric structure of international space law, provides a unique and enforceable mechanism to manage these risks before they materialize, leveraging states’ legal responsibility, control over launch providers, and economic incentives to ensure compliance.

While international treaties are often critiqued as outdated, their allocation of responsibility and liability to states creates strong legal and economic incentives for rigorous ex ante oversight. When states implement rigorous pre-launch scrutiny, ongoing monitoring, and data-sharing obligations, they establish a coherent framework capable of mitigating systemic risks, including those arising from autonomous AI-enabled systems.

Nevertheless, this governance model is not without trade-offs. The precautionary nature of ex ante review may slow the pace of innovation or introduce conservative biases in deployment decisions. Balancing the need for safety against the benefits of technological advancement remains a central challenge. Effective implementation therefore requires not only robust legal and procedural structures but also the careful calibration of oversight thresholds and the integration of technical expertise.

Ultimately, the state-centric approach illustrates that the apparent obsolescence of international space law may be overstated. Instead, it provides a foundational platform upon which contemporary governance of AI-enabled space activities can be built – leveraging historical legal principles, concentrated control points, and interdependent incentives to address the novel risks of the twenty-first-century orbital domain.

1. L. Hall, The History of Space Debris, Space Traffic Management Conference, Embry‑Riddle Aeronautical University, 6, Listopadu, 2014. ↑
2. W. W. Baber & A. Ojala, New Space Era: Characteristics of the New Space Industry Landscape, in Space Business: Emerging Theory and Practice, pp. 3–26, DOI: 10.1007/978‑981‑97‑3430‑6_1, Springer, Singapore, 2024. ↑
3. G. M. Danilenko, International law‑making for outer space, in Space Policy, vol. 37, no. 3, pp. 179–183, DOI: 10.1016/j.spacepol.2016.12.002., Elsevier, Amsterdam, 2016. ↑
4. A. Dey & J. Jagadanandan, Unveiling the realm of AI governance in outer space and its importance in national space policy, in Acta Astronautica, vol. 228, pp. 253–264, DOI: 10.1016/j.actaastro.2024.11.022., Elsevier, Amsterdam, 2025. ↑
5. T. Graham, K. Thangavel & A.‑S. Martin, Navigating AI‑lien Terrain: Legal Liability for Artificial Intelligence in Outer Space, in Acta Astronautica, vol. 217, pp. 197–207, DOI: 10.1016/j.actaastro.2024.01.039., Elsevier, Amsterdam, 2024. ↑
6. L. Hall, The History of Space Debris, in Space Traffic Management Conference, Embry‑Riddle Aeronautical University, 6. Listopadu, 2014. ↑
7. G. M. Danilenko, International law‑making for outer space, in Space Policy, vol. 37, pp. 179–183, DOI: 10.1016/j.spacepol.2016.12.002., Elsevier, Amsterdam, 2016. ↑
8. F. G. Von der Dunk, The Origins of Authorisation: Article VI of the Outer Space Treaty and International Space Law, in F. G. Von der Dunk (ed.), National Space Legislation in Europe: Issues of Implementation, Springer, Dordrecht, 2009. ↑
9. F. G. Von der Dunk, Advanced Introduction to Space Law, in Elgar Advanced Introductions series, Edward Elgar Publishing, Cheltenham, 2020. ↑
10. F. G. Von der Dunk, The Origins of Authorisation: Article VI of the Outer Space Treaty and International Space Law, in F. G. Von der Dunk (ed.), National Space Legislation in Europe, Studies in Space Law, vol. 6, Nijhoff, Leiden, pp. 1–2, 2011; D. Linden, The Impact of National Space Legislation on Private Space Undertakings: Regulatory Competition vs. Harmonization, in Journal of Science Policy & Governance, vol. 8, no. 1, pp. 1–14, 2016. ↑
11. Ibid. ↑
12. C. Stotler, Interpretation as Creation: Article VI of the Outer Space Treaty, in Chicago Journal of International Law, vol. 26, no. 1, pp. 133–146, 2025. ↑
13. F. G. von der Dunk, The Origins of Authorisation: Article VI of the Outer Space Treaty and International Space Law, in F. G. von der Dunk (ed.), National Space Legislation in Europe, Martinus Nijhoff, Leiden, pp. 3–28, 2011; S. E. Bartz, Reframe: A New Interpretive Framework for Non-Governmental Entity Responsibility in Outer Space, in Journal of Air Law and Commerce, vol. 90, no. 2, pp. 154–173, 2025. ↑
14. B. Cheng, Article VI of the 1967 Space Treaty Revisited: “International Responsibility”, “National Activities”, and “The Appropriate State”, in Journal of Space Law, vol. 26, no. 1, pp. 7–33, 1998. ↑
15. F. G. Von der Dunk, The Origins of Authorisation: Article VI of the Outer Space Treaty and International Space Law, in F. G. Von der Dunk (ed.), National Space Legislation in Europe, Studies in Space Law, vol. 6, 2011. ↑
16. J. Handrlica, Kosmické právo správní – skica moderního právního odvětví, in Studia Iuridica Cassoviensia, roč. 13, č. 1, pp. 4–19, DOI: 10.33542/SIC2025-1-02, 2025. ↑
17. Hobe, Schmidt-Tedd and Schrogl, The Cologne Commentary on Space Law, pp. 418–419, Carl Heymanns Verlag, Cologne, 2017. ↑
18. G. Rausser, E. Choi & A. Bayen, Public‑Private Partnerships in Fostering Outer Space Innovations, in Proceedings of the National Academy of Sciences of the United States of America, vol. 120, no. 43, National Academy of Sciences, USA, e2222013120, DOI: 10.1073/pnas.2222013120, 2023. ↑
19. I. Bratu, Mission Impossible? Artificial Intelligence, Space Debris, and the Legal Implications for Space Sustainability, SSRN Electronic Journal, VU University Amsterdam, 28 pp., DOI: 10.2139/ssrn.5112687, 30 Aug 2024. ↑
20. P. D. Cameron, Nuclear Safety After Chernobyl: The Role of International Law, in Leiden Journal of International Law, vol. 1, no. 2, , Leiden, pp. 121–135, DOI: 10.1017/S0922156500000820, 1988.  ↑
21. A. M. Hashmall, After the Fall: A New Framework to Regulate “Too Big to Fail” Non‑Bank Financial Institutions, in New York University Law Review, vol. 85, no. 3, 2010. ↑
22. R. L. Spencer Jr, State Supervision of Space Activity, in Air Force Law Review, roč. 63, str. 75, 127. ISSN 0094-8381, pp. 75-82, 2009. ↑
23. D. O. Cajueiro & V. R. R. Celestino, A Comprehensive Review of Artificial Intelligence Regulation: Weighing Ethical Principles and Innovation, in Journal of Economy and Technology, vol. 4, pp. 77–91, DOI: 10.1016/j.ject.2025.07.001, Elsevier, Amsterdam, 2026. ↑
24. K. Schmalenbach, Convention on International Liability for Damage Caused by Space Objects, in P. Gailhofer et al., Corporate Liability for Transboundary Environemental Harm, Springer, Cham, https://doi.org/10.1007/978-3-031-13264-3_11, pp. 525–530, 2023. ↑
25. J. Hendricks, South Africa’s Place in Space: Towards a New Space Regulatory Framework, in Space Fostering African Societies, pp. 1–24, DOI: 10.1007/978-3-030-89887-8_1, Springer, Cham, 2022. ↑
26. G. Rausser, E. Choi & A. Bayen, Public‑private partnerships in fostering outer space innovations, in Proceedings of the National Academy of Sciences of the United States of America, vol. 120, no. 43, National Academy of Sciences, USA, e2222013120, DOI: 10.1073/pnas.2222013120, 2023. ↑
27. S. Rementeria, Power Dynamics in the Age of Space Commercialisation, in Space Policy, vol. 60, Art. 101472, DOI: 10.1016/j.spacepol.2021.101472, Elsevier, Amsterdam, 2022. ↑
28. Ibid. ↑
29. Ibid. ↑
30. D. O. Cajueiro & V. R. R. Celestino, A Comprehensive Review of Artificial Intelligence Regulation: Weighing Ethical Principles and Innovation, in Journal of Economy and Technology, vol. 4, pp. 77–91, DOI: 10.1016/j.ject.2025.07.001, , Elsevier, Amsterdam, 2026. ↑
31. K. J. D. Chan, G. Papyshev & M. Yarime, Balancing the Tradeoff Between Regulation and Innovation for Artificial Intelligence: An Analysis of Top‑Down Command and Control and Bottom‑Up Self‑Regulatory Approaches, in Technology in Society, vol. 79, Art. 102747, DOI: 10.1016/j.techsoc.2024.102747, Elsevier, Amsterdam, 2024. ↑
32. Ibid. ↑