Governing Joint Innovation
Legal Review and Emerging Technologies under AUKUS Pillar 2
Authors: Natalia Jevglevskaja and Renato Wolf
Introduction
Although public attention since the September 2021 announcement of the AUKUS partnership[1] between Australia, the United Kingdom (UK) and the United States has largely (US) centred on nuclear-powered submarines,[2] there is broad agreement that the development and sharing of advanced capabilities under Pillar 2 will be equally central to the partnership’s success.[3] Driven in significant part by China’s rapid and unprecedented military modernisation – including in areas such as artificial intelligence, hypersonic systems and electronic warfare – the Indo-Pacific is experiencing accelerated technological change in the defence domain. Against this backdrop, it is essential that the AUKUS partners retain a technological edge in order to sustain credible deterrence and contribute to regional stability.[4] Pillar 2 responds to this imperative by establishing a trilateral capability accelerator that targets technologies of immediate strategic relevance, linking government, industry and research communities across the three States to enhance interoperability and promote collaborative innovation.[5]
The apparently intensifying pace of advanced capability development under Pillar 2 creates a parallel imperative: ensuring that relevant governance frameworks – including legal review processes – evolve at a comparable pace. Certain Pillar 2 technologies – most notably AI-enabled systems – are not static artefacts but adaptive systems whose behaviour may change over time through software updates, data inputs and integration with other (weapon) systems and platforms. This dynamism has already strained national legal review frameworks, which were largely designed for stable, predictable technologies, prompting ongoing efforts to recalibrate review methodologies.[6] Other Pillar 2 technologies – such as quantum capabilities and advanced electronic warfare systems – are likely to test these frameworks in comparable, if not more profound, ways, further exposing the limits of existing approaches.
Crucially, Pillar 2 requires policymakers and legal review authorities to think one step ahead. If capabilities developed under Pillar 2 are intended to be designed, tested and deployed jointly, it is hard to see how this can be meaningfully achieved without triangulating legal considerations relevant for each AUKUS partner. In fact, this has prompted at least one analyst to call for ‘multinational’ – that is, AUKUS-specific – legal review processes.[7] While such proposals have intuitive appeal, the more difficult question is what form such a review mechanism would take, how it could operate in practice as a single process, and whether it would be viable in the first place given that the three AUKUS partners might be bound by different international law obligations in relation to capability development and also operate under distinct – and at times more restrictive – national legal and policy frameworks.
Despite its significance, the implications of AUKUS Pillar 2 for the legal and policy requirements to conduct legal review of military capabilities[8] have received little, if any, sustained attention in the expert literature. Addressing this gap, the paper critically examines the frictions arising from the divergent approaches to legal reviews across the three AUKUS partners and explores the viability and limits of possible mitigation measures. Specifically, we first discuss the feasibility of a joint legal review mechanism where the capabilities developed jointly are also reviewed jointly. We then examine an alternative measure – a trilateral legal review coordination mechanism – where the national review processes can be coordinated, results shared and problems addressed. The aim of proposed mitigation measures is to support the objectives of Pillar 2 and identify key legal, policy and practical challenges that such mechanisms would need to address. For the Australian Army, the issue is of particular relevance as many emerging capabilities – especially AI-enabled systems – may be developed in the framework of Pillar 2, and issues with legal review may thus particularly affect the Australian Army. The purpose of the paper is therefore to draw attention to an issue of growing importance for Australia and its AUKUS partners and to stimulate more systematic discussion of how legal review processes should evolve in the context of jointly developed emerging technologies.
In what follows, the paper undertakes a stocktaking exercise of AUKUS Pillar II, assessing the extent to which its ambitious commitments to advanced capability development and industrial integration have translated into tangible outcomes. The paper outlines the national legal review processes in Australia, the UK and the US and identifies key areas of convergence and divergence. Later in the paper we examine the implications of these similarities and differences for the feasibility of a joint legal review mechanism, taking into account the broader practical challenges currently confronting Pillar 2. In view of significant legal and practical hurdles to the establishment and maintenance of a joint mechanism, the paper outlines key aspects of a more likely arrangement that AUKUS partners might pursue – a trilateral legal review coordination mechanism.
AUKUS Pillar 2. Nearly Five Years In: Where Are We Now?
The creation of AUKUS was driven by the recognition ‘that the Indo-Pacific is entering a period of unprecedented strategic volatility’.[9] While Pillar 1of the partnership seeks to enable the acquisition of nuclear powered submarines by Australia, Pillar 2 comprises ‘a suite of cooperative activities’ through which the United States, the United Kingdom, and Australia seek to ‘develop and field “advanced capabilities”’.[10]
Initially, the AUKUS leaders announced four areas of cooperation in critical technologies, namely ‘cyber capabilities, artificial intelligence, quantum technologies, and additional undersea capabilities’.[11] However, not long thereafter, in April 2022, these were expanded to include two further technology-focused areas (hypersonic and counter-hypersonic capabilities and electronic warfare capabilities) and two functional areas, namely innovation and information sharing.[12] The three States have been coordinating Pillar 2 activities through various mechanisms, including working groups (also known as ‘workstreams’).[13]
The technological breadth of the undertaking is extraordinary. Unlike Pillar 1, centred on nuclear-powered submarines, Pillar II encompasses a broad, evolving suite of advanced and emerging technologies. Each of the listed technological domains represents a complex ecosystem involving software, hardware, data infrastructure, testing environments, and operational doctrine. While some capabilities sit within specialised domains, such as undersea systems and hypersonic technologies, others – most notably AI, quantum technologies, and cyber capabilities – are cross-cutting in nature. With applications extending across military and civilian sectors, they are likely to define the trajectory of warfare and security at national, regional, and global levels well into the future. Sustained joint technological progress will therefore be essential for the AUKUS partners to outpace strategic competitors in this field, notably key Indo-Pacific actors such as China and India.[14]
The timing of Pillar 2 appears particularly opportune. Analysis by the Australian Strategic Policy Institute indicates that China currently leads in high-impact research in 19 of 23 critical technologies, with a commanding position in areas such as hypersonics, electronic warfare and key undersea capabilities.[15] Still, the picture changes when the collective technological capacity of the AUKUS partners is considered. In several strategically important fields – including ‘autonomous systems operation technology, advanced robotics, adversarial AI reverse-engineering and protective cyber’ – the combined strength of Australia, the UK and the US places them at the forefront globally.[16] These developments take place against the backdrop of China’s innovation-driven development strategy, launched in 2016,[17] which aims to establish the country as a global leader in science and technology by 2050.[18] Pillar 2 thus seeks to steer its members’ efforts in counterbalancing China’s rivalry for technological hegemony.
Despite the central role attributed to Pillar 2 in the partnership’s long-term strategic vision, publicly available information about its practical implementation remains limited. As commentators have observed, access to the AUKUS material remains scarce not only because programs such as the SSN-AUKUS submarines will take decades to deliver, but also because both Pillar 1and Pillar 2 involve highly sensitive technologies whose disclosure is tightly controlled.[19] That said, calls for the creation of a ‘Pillar 3’ – variously proposed to address issues such as critical minerals[20] or space capabilities[21] – have been voiced since 2024, even as the operational contours of Pillar 2 itself remain only partially understood.
The enabling architecture for Pillar 2 is still evolving. Important legislative and regulatory steps have been taken, including recent adjustments to export control frameworks in all three States.[22] Yet assessments of progress remain mixed. While some commentators defend the significance of these reforms,[23] others point to persistent structural obstacles, particularly in areas such as the US Foreign Military Sales system, which is regarded as slow, rigid and ill-suited to collaborative development of rapidly evolving technologies such as AI or quantum systems.[24] More broadly, analysts note that Pillar 2 has struggled to generate tangible outcomes and faces pressure to demonstrate clearer priorities. [25]Nevertheless, even if the scope of Pillar 2 is eventually narrowed or refocused, its central objective – enhanced interoperability through the joint development of advanced capabilities – will remain fundamental. This enduring emphasis on collaborative capability development reinforces the importance of addressing the legal and governance questions examined in this paper, particularly if Pillar 2 expands in the future to include additional partners with differently structured legal review mechanisms, such as New Zealand or Canada, or partners for which no comparable mechanisms are publicly known, such as Japan or South Korea.[26]
National Legal Reviews: Points of Convergence and Divergence
A central issue for the AUKUS partners is whether and how the ambitions of Pillar 2 can be pursued consistently with their obligations under international law and other applicable domestic law and policy commitments. Although the expert literature suggests that international law places relatively few restrictions on the inter-State sharing of advanced technologies of the kind envisaged under Pillar 2,[27] the joint design and development of military capabilities engage a range of international legal obligations. Among others, all three AUKUS States must ensure compliance with the law of armed conflict (LOAC), which applies to their cooperative activities in this area.
One of the core principles of LOAC, set out in Article 35(1) of Additional Protocol I to the Geneva Conventions (AP I / Protocol), is that the right of belligerents to choose weapons, means, and methods of warfare is not unlimited.[28] Article 36 of the Protocol gives practical effect to this principle by requiring States, prior to the development or acquisition of a military capability, to assess whether it can be used consistently with international law.[29] Described commonly as the duty to conduct ‘weapons reviews’ or ‘legal reviews’ (for the further: legal reviews) this requirement may be tightened by States to incorporate more restrictive obligations under domestic law or policy.[30]
In what follows, we briefly outline domestic legal review regimes in the three States (section C1). These regimes display both commonalities and points of divergence in the framing and conduct of legal reviews (section C2). While areas of convergence may facilitate the development of a joint approach under AUKUS, differences in scope, institutional design, or review practice may signal potential pressure points and practical hurdles for any such future cooperation.
Legal Reviews: Implementation in Australia, the United Kingdom and the United States
Australia and the UK are both parties to AP I and each conducts legal reviews pursuant to their obligations under Article 36 of the Protocol. The United States, by contrast, has never ratified this instrument, although it is a signatory. Notably, the United States’ practice of reviewing weapons predates the Protocol’s adoption in 1977, with documented examples of such reviews as early as 1918, undertaken as part of an informal review practice.[31] The US conducts weapons reviews as a matter of policy treating them as a best practice for implementing customary and treaty-based constraints on weapons use.[32]
Australia
Australia formally assumed the obligation to conduct legal reviews in June 1991 upon ratifying AP I. For more than a decade, legal reviews were performed on an ad hoc basis, that is without a structured institutional framework. This changed in June 2005, when the Australian Defence Force (ADF) formalised the review process through a Defence Instruction which established a consistent procedure within the ADF by laying out roles and responsibilities for the conduct of legal reviews.[33]
The directive was subsequently revised on several occasions. In May 2020, the process was updated to ensure closer alignment with Defence’s capability acquisition lifecycle. As part of this revision, legal advice under Article 36 became integrated into high-level decisions, including early-stage consideration of whether proposed capabilities should proceed to procurement.[34]
Another series of reviews were undertaken between 2022 and 2023 and resulted in the adoption of the Guide to the Legal Review of New Weapons, Means or Methods of Warfare (ADF Guide), which remains the legal basis for conducting reviews in Australia at the time of writing.[35] The Guide is organised into three parts. Part 1 sets out key definitions, procedures, and responsibilities. Overall responsibility for Article 36 reviews rests with the Director-General of the Military Legal Service, while assessments are conducted in practice by the Directorate of Operations and International Law (DOIL) at the request of authorised Defence entities. Capability managers initiating a review are required to complete an online legal review report form and provide supporting technical documentation.[36] DOIL then conducts the assessment based on the submitted material. Each review is undertaken by two legal officers: a primary reviewer and a senior reviewer exercising delegated authority. The completed assessment is formally counter-signed by the requester. Part 2 of the Guide outlines Australia’s established interpretation of the scope of Article 36, and Part 3 addresses additional considerations relevant to the assessment of autonomous weapon systems (AWS).
United Kingdom
The UK ratified AP I in January 1998. Although legal reviews of military capabilities had been conducted before that date on an informal basis, ratification prompted the establishment of a structured process for legal reviews. Since then, the process has significantly evolved in response to developments in military technology and procurement practice.[37]
Today, legal reviews are conducted on behalf of the Ministry of Defence by a tri-service team of military legal advisers within Defence Futures and Force Design (F&FD), previously known as Defence Futures (DF) and, before July 2024, the Development, Concepts and Doctrine Centre (DCDC).[38] Having one legal officer from each Service ensures that all Services remain actively engaged in the discharge of the UK’s Article 36 obligations.
The review process is governed by a Joint Service Publication which is not in the public domain.[39] The Publication sets out the applicable procedures and includes a template that capability owners must complete in order to initiate a review.[40] Once the request is received, the reviewing authority develops an assessment plan and requests relevant information at the earliest opportunity. Since the information required varies from one capability to the next, the team consults with appropriate subject-matter experts to develop the necessary technical and evidentiary basis for the assessment. Final advice is typically delivered before the capability is introduced into the UK Armed Forces’ force structure with interim advice governing capability development where relevant.[41]
United States
While, as mentioned, the US is not formally bound by the AP I, legal reviews of weapons have been a formal policy requirement since 1974.[42] The responsibility for conducting these reviews is closely linked to acquisition authority within the Department of Defense (DoD). The DoD component responsible for acquiring or developing a weapon system is also responsible for conducting its legal review. Accordingly, military services – the Army, Navy, and Air Force, as well as certain Combatant Commands – including Cyber Command and Special Operations Command – review the legality of the weapons they acquire or develop.[43]
While procedural steps vary slightly among the services, there are no substantial differences in the standards applied.[44] Each review results in a written legal opinion. Draft opinions are typically circulated to the relevant Project Office to confirm technical specifications and ensure appropriate classification markings. The draft is usually shared with other services for concurrence; however, such consultation does not remove the requirement for each service to conduct its own review.[45]
Points of Convergence and Divergence
It may appear intuitive that capabilities developed jointly under AUKUS Pillar 2 should also be reviewed jointly. Michael Sharpe, Chief Executive Officer of the AUKUS Forum,[46] has observed that he likes to think of AUKUS as ‘one big factory across three nations.’ [47] From a legal review perspective, such an integrated model of capability development may equally call for a single joint legal review across all three jurisdictions. But is this feasible? This section focuses on the points of convergence and divergence in the practice of legal reviews, in order to assess whether the future establishment and operation of ‘a joint legal review mechanism under AUKUS Pillar 2’ may be facilitated or constrained.
Points of Convergence
To begin with, in all three AUKUS States, the authority to conduct legal reviews is vested within their respective departments of defence and carried out by subordinate units within those institutions. The UK’s Defence Futures and Force Design (F&FD) is an organisational unit within the Ministry of Defence (responsible inter alia for future capability planning and concept development) and review authorities both in Australia and the US (DOIL and individual services[48]) equally sit within respective Departments of Defence. We believe that the embeddedness of legal reviews within the defence establishment itself, suggests a broadly similar institutional setting, culture and mode of operation – an aspect likely to facilitate the development of a joint review mechanism.
Contrast this with the Swedish legal review model, where the Delegation for International Law Monitoring of Arms Projects, established by government ordinance in 1974, operates as an independent and multidisciplinary body formally separate from both the government and the armed forces.[49] It appears that the institutional setting in the AUKUS States is likely to suggest more comparable working practices and professional outlooks, as opposed to approaches where the legal review function is assigned to an independent body outside the defence establishment. An external review authority may operate with greater structural independence from capability development processes and also follow different procedural preferences and rhythms.
Second, in all three States, legal reviews function as a component within a broader governance architecture designed to ensure that new capabilities can be employed consistently with international law, in particular IHL.[50] Within this broader governance structure, the legal review operates alongside a range of other legal, technical, and operational processes that collectively ensure compliance with applicable law. It plays a specific preventive role: it assesses whether a military capability is capable of being used lawfully when it is introduced into service.[51] Ensuring that the weapon is indeed used lawfully in practice depends on the wider institutional framework governing military operations. Australia explicitly articulated this position in 2019 in its ‘system of control’ governing the military use of force.[52] Accordingly, the legal review process forms one (albeit important) component of a layered framework covering ‘all aspects of a weapon system from design through to engagement,’[53] including weapon design, testing, evaluation and review, training and certification, rules of engagement, and command oversight.
The UK and the US positions are similar. Officials responsible for legal reviews in the UK emphasise that they constitute an important component of the UK’s international law compliance processes but are only one element within a broader defence framework aimed at ensuring the lawful development and deployment of military capabilities.[54] The US equally lists legal reviews as one of the requirements within the broader framework of compliance with the laws of war.[55]
The fact that all three States conceptualise legal reviews as one element within a broader system of the capability development and use reflects a shared governance philosophy. This common understanding of the role of legal reviews within capability acquisition is, in our opinion, likely to facilitate coordination between the three systems, even if the precise institutional and governance arrangements differ.
Third, legal review processes in the three AUKUS States have evolved continuously in response to both technological developments and changes in defence capability procurement systems. In particular, all three States in the past decade turned their attention to the challenges posed by AI capabilities and increasingly complex processes of human–machine interaction, reflecting a shared recognition that traditional legal review procedures require recalibration to address software-driven systems whose operational behaviour may change over time. The three States continue to explore ways of integrating legal reviews more closely into capability acquisition cycles so that legal advice can inform design, testing, and development decisions at earlier stages.[56] This convergence in regulatory focus, timing, and appreciation of the challenges ahead creates favourable conditions for the development of a joint review mechanism, as the participating States share a common understanding of the legal and operational implications of emerging technologies and the role of legal reviews in addressing them.
Points of Divergence
Whilst certain aspects of the national legal review processes suggest feasibility of a joint review mechanism, important points of divergence between national systems may pull in the opposite direction.
The scope of legal reviews
To begin with, important differences arise in the way Australia and the UK define the scope of legal reviews compared to the US. Ultimately, the US’s narrower approach creates uncertainty as to what would be reviewed jointly and risks excluding capabilities that Australia and the UK would otherwise review. Our reasoning is as follows.
Both States explicitly frame the scope of their legal review obligation as covering ‘weapons’ and ‘means of warfare’, as per Article 36 AP I. Australia’s Guide defines ‘a weapon’ broadly as any ‘arms, munition, device, whether tangible or intangible, designed or intended to be used to cause: (1) injury to, or death of, persons; or (2) damage to, destruction of or neutralisation of objects’ and explains that ‘means of warfare’ include ‘weapons as well as weapon systems, which includes all related equipment, materials, and delivery mechanisms necessary for a weapon to function as intended’.[57] The UK practice likewise adopts an expansive interpretation of ‘means of warfare.’ Earlier UK guidance indicated that Article 36 reviews encompass ‘weapons in the broader sense of that word’ as well as ‘sighting equipment, laser designators, target acquisition equipment, data links and software used for processing targeting data’, and even capabilities such as tactical radios or navigation equipment may be submitted for review before a determination is made that a formal assessment is unnecessary.[58]
By contrast, the US formally focuses on the review of ‘weapons’ and ‘weapon systems.’ Definitions across the US services generally describe weapons as devices intended to injure or kill persons or damage or destroy property.[59] This terminology may appear narrower because US approach does not explicitly refer to ‘means of warfare.’ However, some US definitions of weapons and weapon systems are so expansive – covering a wide range of components and equipment required for their operation – that, in practice, the scope of review may capture much, if not all, of what other two States describe as ‘means of warfare’. Consider, for example, the US Air Force definition of a ‘weapon system’ which includes ‘a combination of one or more weapons with all related equipment, materials, services, personnel, and means of delivery and deployment (if applicable) required for self-sufficiency’.[60]
The divergence is, however, more profound for launch or delivery platforms. The US services explicitly exclude such platforms – for example, ships, aircraft, or armoured vehicles – from the definition of weapons subject to review.[61] By contrast, the UK and Australian approaches do not appear to exclude such systems a priori. Indeed, Australia’s definition of ‘means of warfare’ refers to the ‘delivery mechanisms necessary for a weapon to function as intended’, suggesting that the scope of review may extend more readily to the platforms that enable the use of force. Likewise, nothing in the UK’s broad conceptualisation of weapons and means appears, at least prima facie, to exclude such platforms from consideration.
Another divergence arises with respect to ‘methods of warfare.’ Australia defines methods broadly as ‘the manner in which weapons and certain capabilities are used in warfare to harm, destroy, or neutralise the enemy’,[62] and UK practice likewise considers methods to extend beyond ‘normal’ or ‘anticipated’ uses of weapons to ‘ways in which warfare is conducted’.[63] US services, in comparison, generally assess the ‘concept of employment’ aka ‘intended use’ of a weapon, rather than reviewing methods of warfare as such.[64] At first glance, this approach appears narrower. In the US practice, however, broader methods of warfare (including, for example, tactics, techniques and procedures) are not ignored but may instead be addressed through separate policy, doctrinal, and operational oversight processes outside the formal legal review regime.[65] In contrast, in the UK, the inclusion of a wider range of methods in the legal review framework reflects the institutional remit of F&FD, whose focus on strategic foresight, concepts, capability strategy, and force design analysis necessarily engages with a broader issue of methods employed in warfighting.
The institutional separation reflected in the US treatment of broader methods of warfare – compared to the more integrated approaches of Australia and the UK – may complicate the design of a joint AUKUS review mechanism and weaken its prospects. At a minimum, it introduces uncertainty as to the appropriate scope of any joint review and the timing of assessments within the capability lifecycle. Aligning these approaches would require agreement on whether such methods should be incorporated into a single review process or continue to be addressed through parallel mechanisms, thereby adding layers of coordination and potential friction.
Standard of review: Superfluous Injury and Unnecessary Suffering
Another point of divergence concerns how the three States apply the rule prohibiting the infliction of superfluous injury or unnecessary suffering (SI/US) in their review processes.[66] While all three States largely follow the same two-step methodology – examining first compliance with specific prohibitions or restrictions under treaty or customary IHL and then with general prohibitions or restrictions under treaty and custom, including whether the weapon is capable of being used in accordance with the principle of distinction and whether it causes SI/US – differences arise in how the latter standard is interpreted and applied.
There is broad agreement on the substantive purpose of the rule. As the International Court of Justice explained in the Nuclear Weapons Advisory Opinion, the prohibition concerns weapons that cause ‘harm greater than that unavoidable to achieve legitimate military objectives’.[67] The rule therefore requires a balancing of the humanitarian harm caused by the weapon against the military advantage expected from its use. The central (and contentious) issue for present purposes is what gives rise to the harm – whether it is the inherent nature of the weapon or the intention of its designer.
Australia and the UK apply the standard codified in Article 35(2) of AP I, which prohibits the use of weapons ‘of a nature to cause’ SI/US. Under this formulation, the legality of a weapon depends primarily on its inherent characteristics and effects.
Although strong arguments suggest that the customary IHL test is that ‘of a nature’,[68] the US has long taken the position that the correct formulation of the rule is that weapons must not be ‘calculated to cause superfluous injury’. This approach, first adopted in the 1956 U.S. Army Field Manual on Land Warfare,[69] reinforced since by prominent US commentators[70] and currently reaffirmed in the U.S. Department of Defense Law of War Manual,[71] appears to focus on the purpose or design intention behind the weapon. In the US view, the phrase ‘calculated to cause’ more accurately reflects the intent of the original French text of the 1907 Hague Regulations.[72]
The difference in tests applied is not merely theoretical. Consider an AI-enabled automated sentry weapon designed to neutralise approaching enemy personnel. The system uses computer vision to identify human movement and is programmed to fire repeated low-calibre rounds at a detected combatant until the algorithm determines that the person is no longer mobile. The design objective is operationally framed as ensuring reliable neutralisation in high-tempo environments where human operators may not be able to intervene quickly.
Under the US ‘calculated to cause’ SI/US formulation, the relevant question is whether the weapon was designed with the purpose of increasing suffering beyond what is required by military necessity. In this case, it could be plausibly argued that the system is intended simply to ensure neutralisation of the target, not to increase suffering. The repeated firing pattern arises from the algorithmic logic used to confirm incapacitation, not from an intention to aggravate injury.
However, under the ‘of a nature to cause’ SI/US standard applied by Australia and the UK pursuant to Article 35(2) AP I, the analysis would focus on the predictable effects of the weapon in operation.[73] A system that systematically continues to fire at already incapacitated personnel until the algorithm registers immobilisation could foreseeably produce multiple and unnecessary wounds beyond those required to neutralise the combatant. Because other anti-personnel weapons – for example, those used by human sentry personnel – can achieve the same military purpose (incapacitation) without predictably causing such cumulative injury, the system could be considered ‘of a nature’ to cause unnecessary suffering.
Given these differences in application, any joint legal review mechanism would, at a minimum, need to address explicitly the competing formulations of the rule and determine how to proceed where those differences lead to divergent outcomes – such that the same capability may be permissible for the US but not for the UK and Australia.
Standard of review: Environmental damage
A further point of divergence concerns the protection of the natural environment in armed conflict. Australia and the UK are bound by Art. 35(3) AP I, which prohibits the employment of methods or means of warfare that are intended, or may be expected, to cause ‘widespread, long-term and severe damage to the natural environment’. Although there is now sufficient State practice to support the proposition that this rule has crystallised to customary international law,[74] the US has positioned itself as a persistent objector to it and does not include the rule in its standard of legal review.[75]
This divergence in approaches suggests that a capability developed jointly under AUKUS Pillar 2 might be assessed against different environmental standards by the participating States. Consider a long-endurance autonomous underwater system designed to deploy large numbers of smart naval mines in sensitive maritime environments or an AI-enabled autonomous area-denial systems designed to ignite large-scale vegetation fires in order to restrict enemy movement. If the expected environmental effects of such systems included a widespread, long-term and severe disruption of affected marine ecosystems in the former case and a widespread, long-term and severe degradation of vegetation, soil and wildlife habitats in the latter, the systems would be impermissible under the standard applied by the UK and Australia but likely permissible under that applied by the US.
This issue is not one of simple arithmetic where the position of two States would automatically prevail over that of one. The mere existence of a joint arrangement would not, without more, require the US to revise its settled interpretations of international law or assume obligations it has consistently declined to recognise. In practice, however, this could create a structural difficulty for jointly developed capabilities under Pillar 2, as joint funds might be invested in a capability that ultimately can be lawfully employed by only one partner.
One possible response would be for Australia and the UK to insist, at an early stage of capability development, that systems likely to meet the environmental damage threshold should not be pursued within the Pillar 2 framework. However, apart from the US expected objections to that as a matter of law, such an approach also raises practical and political concerns. Predicting environmental effects decades in advance – particularly for complex, software-enabled systems – may be difficult, and imposing ex ante constraints could limit technological experimentation or innovation. The feasibility of such an approach would therefore depend on whether the partners are willing to accept stricter common standards for jointly developed systems or, alternatively, to tolerate the possibility that certain capabilities will ultimately be usable by only some or one of them.
Standard of review: international law vs additional legal considerations
Differences in States’ international legal obligations, together with the inclusion of domestic law (and policy) in national legal review processes could pose yet another significant hurdle for the establishment of a joint legal review mechanism. In the US, reviews conducted by the military services explicitly assess whether a weapon / weapon system complies not only with international law but also with domestic law and policy.[76] Likewise, although the primary focus of Australia’s Article 36 review is compliance with international law, the Australian framework recognises that issues arising under domestic law may surface and require attention during the review process.[77] The UK similarly frames legal reviews primarily in terms of international law obligations, although in practice it would be difficult for reviewing authorities to disregard relevant domestic legal constraints.[78]
This issue is significant because domestic legal frameworks differ substantially between States and may impose additional or distinct constraints on the development, acquisition or use of military capabilities. While a State cannot invoke its domestic law to avoid its obligations under international law, nor require another State to comply with its national law,[79] these differences nonetheless complicate the operation of a joint review mechanism. The concern here is that more restrictive domestic standards applied by one partner could, in effect, be imposed on others through such a mechanism, thereby bypassing ordinary domestic legislative or regulatory processes. However, even if the AUKUS partners were to amend their domestic legal and regulatory frameworks to accommodate restrictions applied by other partners within a joint review mechanism – that is by creating an explicit legal basis for incorporating more restrictive standards applied by any of the AUKUS partners – it is difficult to imagine that the three AUKUS States would go down that path. After all, a mechanism designed to produce a single determination of legality acceptable to all three partners could prohibit or restrict a capability that would otherwise be lawful under the applicable rules of international law. In circumstances where a State would have been free to develop and employ the same capability if acting independently, such an outcome would be difficult to justify.
Any joint mechanism would therefore need to clearly define how to address situations in which participating States are bound by differing international or domestic legal obligations, or where they adopt divergent interpretations of shared legal standards. In that regard, a number of Workshop participants expressed concern that a joint review mechanism could reduce or dilute the applicable legal standard, ultimately rending it little more than a fig leaf. They pointed to what they termed the ‘least common denominator’ problem, which could manifest in two ways.
First, if one categorises the relevant law into four groups – Category A (international and domestic law binding on Australia), Category B (international and domestic law binding on the UK), Category C (international and domestic law binding on the US), and Category D (treaty and customary IHL rules binding on all three) – it might initially seem that cooperation under AUKUS Pillar II could extend only to capabilities falling within Category D, and not to those affected by Categories A, B, or C.
Consider, for example, the Convention on Cluster Munitions:[80] the US is not a party, whereas Australia and the United Kingdom are. On this logic, Category D would exclude obligations arising under that Convention. The concern, put differently, is that the collective baseline would reflect only the lowest common set of shared obligations, thereby reducing the level of scrutiny applied to a capability under development. In the example above, this raises the risk that, within a joint legal review framework, obligations under the Convention could be sidelined because they do not bind all three partners.
Another way to frame the ‘least common denominator’ problem is as a selective, issue-by-issue approach in which, for each legally relevant aspect of a capability, the most permissive standard among the three States is applied. Thus, if the US adopts the least restrictive interpretation of unnecessary suffering (see our discussion in section 2 ii) b) above) that standard would govern that element of the joint review; if Australia applies the least restrictive approach to indiscriminate weapons, its standard would govern that element, and so on. The likely consequence is that the joint review would deem a capability lawful for all three States, even though particular features of that capability would fail to meet the stricter standards each State applies in its own system. In effect, the process would aggregate permissive elements across jurisdictions, resulting in a systematic lowering of the overall level of legal scrutiny.
We suggest, however, that this is not how practice is likely to operate. Each of the three States has thus far conscientiously reviewed weapons against the international and domestic legal obligations binding upon it. It is difficult to imagine that they would disregard applicable law merely because a capability is being developed jointly under AUKUS Pillar II. Rather, reviewing authorities in each State are likely to identify the international and domestic rules applicable to them and to specify any developmental, acquisition, or operational limitations required by those obligations. For example, where a capability engages obligations binding on Australia or the UK but not the US, those States may approve participation subject to conditions – such as restrictions on design features, targeting parameters, or permissible operational use – while the US conducts its review against its own set of legal obligations and interpretations. In this way, divergence in applicable law is managed through nationally tailored caveats rather than by lowering the overall standard of review. It is for these reasons that we consider a Trilateral Coordination Mechanism to be the most appropriate framework for managing legal review processes in respect of AUKUS Pillar II capabilities (see section E below).
One final remark is appropriate here. A joint legal review that genuinely answers the question of whether a capability may be used lawfully by each State would require precisely the opposite logic to that feared under a ‘least common denominator’ approach. In theory, it cannot be discounted, that rather than defaulting to the most permissive rule, a joint legal review would apply, for each legally relevant aspect, the most restrictive standard among the participating States. If Australia adopts the most stringent position on unnecessary suffering, that standard would need to govern that element of the assessment; if the US applies the strictest approach to indiscriminate weapons, its standard would control that element, and so on. The practical consequence is that any capability capable of passing such a composite review would, by design, reflect the highest level of constraint across the three systems. The difficulty of such is approach is, however, that each state will likely have restrictions on the use of a given capability they would not otherwise have if they only applied their own legal standards. While the approach may be lawful, it thus likely limits the overall military utility of a given capability..
One Joint Legal Review: Meaningful and Feasible or Rather Not?
As the preceding discussion illustrates, in some respects, the ambition of ‘one joint legal review’ is supported by existing convergence between the national practices of legal reviews. As shown in the preceding section, these review mechanisms are embedded within defence institutions, operate as part of broader capability acquisition frameworks, and have evolved in parallel in response to challenges posed by emerging technologies, particularly AI-enabled systems. This shared institutional setting, the comparable functional and structural role of legal reviews, and a common appreciation of the challenges posed by cutting edge and emergent technologies – reflected in a concerted effort to address them – provide a plausible foundation for developing a joint review mechanism.
However, important deviations cast doubt on the viability and utility of such a mechanism. Divergence in the scope of legal reviews is only the first from a range of stumbling blocks to consider. While Pillar 2 identifies six technology-focused areas, not every technology falls in the scope of the legal review. As follows from the section C(2)(ii) above, all AUKUS partners conduct legal reviews of capabilities that, in their normal or intended use, are designed to inflict bodily harm or cause physical damage.[81] As a practical matter, therefore, where a Pillar II technology is integrated into a capability that is expected, in its ordinary employment, to result in damage, destruction, injury, or loss of life, that capability falls within the scope of legal review. For example, where an AI component is integrated into a weapon system to improve navigation or target identification and selection, that component falls within the scope of the legal assessment as part of the overall system.
Still, differences in the scope of review – most notably the US narrower approach to ‘methods of warfare’ and its exclusion of certain ‘means of warfare’ (such as platforms) – create uncertainty as to what capabilities are likely to be reviewed jointly. Most notably, such different scoping raises the risk that capabilities which the UK and Australia would ordinarily review within their legal review frameworks might, if developed jointly, fall outside the review ambit. Where the US addresses certain methods of warfare outside the formal legal review process, agreement on the appropriate alternative institutional locus and timing of such assessments would be required, adding complexity and creating potential gaps in oversight.
Notably, even within a single State, defining and classifying emerging military capabilities in a way that clearly signals to acquisition authorities which systems require legal review can be challenging. For example, the US Army reviews cyber weapons, a term that includes certain cyber capabilities,[82] while the US Air Force prefers the broader term cyber capabilities, which may not necessarily be classified as weapons.[83] At the domestic level, this inconsistency in terminology could create confusion for an acquisition authority supplying the same cyber capability to both services. In a trilateral AUKUS context, similar terminological divergences – rooted in established national practices – could generate further ambiguity, complicating the agreement on what must be reviewed.
Further, divergences in the applicable legal standards, including the interpretation of SI/US rule and the treatment of environmental harm, may lead to different outcomes for the same capability. Last but not least, the fact that the three partners may be bound by different international and domestic law obligations and follow distinct interpretations of common standards presents a significant structural obstacle: any joint mechanism would struggle to produce outcomes that are simultaneously legally valid and operationally usable for all three States without either disregarding constraints applicable to some States or forcing a degree of alignment that States are neither legally required nor politically willing to accept.
A further structural constraint is the absence of a clearly defined institutional platform to anchor a joint legal review mechanism. The legal review mechanism would likely need to be embedded within a stable trilateral setting capable of consolidating expertise and ensuring continuity across projects. Restated, if Pillar 2 is indeed conceived as ‘one big factory across three nations,’ as Sharpe suggests, such a factory would necessarily require a dedicated office for a review authority within its premises. At present, however, no such structure exists. AUKUS Pillar 2 remains characterised by fragmented governance: coordination occurs through evolving working groups, while capability development and funding decisions continue to be made largely through national processes.[84] This results in projects advancing on parallel, rather than truly integrated, tracks, with limited alignment of objectives, timelines, standards, and oversight practices. The problem is compounded by the lack of dedicated funding streams for Pillar 2 activities.[85] Rather than being supported by a distinct budget, Pillar 2 related work is absorbed into existing national programmes, reducing visibility, predictability, and strategic coherence. Today, these institutional and financial gaps make it difficult to envisage how a joint legal review mechanism could be effectively established, resourced, and sustained over time.
More fundamentally, the establishment of such a mechanism raises questions as to whether AUKUS States can discharge their legal review obligations through a mechanism not managed by their own authorities. While domestic review processes may, in practice, include exchange officers from partner States, ultimate responsibility remains with the reviewing State. In a joint mechanism, however, the allocation of responsibility would be less straightforward and would need to be expressly addressed. Particularly in case where the joint mechanism might be expected to take into account domestic prohibitions or restrictions applicable in one partner State that do not otherwise bind the others.[86] In such circumstances, participation in a joint review process would likely require enabling legislation – ie parliamentary authorisation – to address both the issue of responsibility and the incorporation of foreign domestic constraints. While this matter neither needs to be addressed in more detail nor resolved here, it underscores further legal and practical hurdles to the implementation of a joint review mechanism.
Lastly, even if a joint review mechanism were established, there remains a risk that one partner –most notably the US – could dominate its operation. A trilateral mechanism would need to function on the basis of equality, ensuring that each State retains an equal voice and that decision-making authority is balanced. Allowing one State to act as the de facto ultimate reviewer would risk reproducing asymmetries already evident in the export control domain, where the United States effectively set the parameters to which Australia and the United Kingdom adapted. Specifically, rather than developing a genuinely trilateral export control framework tailored to the objectives of AUKUS Pillar 2, the United States effectively insisted on the primacy of its domestic approach, long criticised by experts for its shortcomings.[87] Through the National Defense Authorization Act for Fiscal Year 2024 (NDAA), the U.S. Congress created a national exemption from certain U.S. defence export licensing requirements for Australia and the United Kingdom, conditional upon the U.S. President certifying to Congress that both countries have implemented export control systems comparable to those of the United States.[88]
Combined, these factors suggest that while a joint review mechanism may be conceptually appealing, its practical design and operation would be fraught with complexity and may ultimately require a more flexible, coordinated – rather than unified – approach. We discuss this approach in the next section.
Trilateral Legal Review Coordination Mechanism
While a joint legal review mechanism poses significant legal and practical challenges, relying on domestic review processes alone is equally unsatisfactory. As set out above, even though the AUKUS partners share a common core in the scope of legal reviews – covering weapons and their use – they diverge in the treatment of certain means and methods of warfare, and – in respect to the same military capability – may be bound by different legal obligations or adopt divergent interpretations of common norms.
A more feasible alternative may therefore lie in the establishment of a trilateral coordination mechanism (TCM). Such a mechanism – whether formal or informal – could serve as a structured forum for addressing issues arising in the context of national legal reviews. In particular, it could enable regular exchange of information and facilitate coordination on key aspects of the review processes and capability development. We suggest that the following key aspects be considered in the establishment of a TCM.
First, given that personnel responsible for conducting legal reviews in each AUKUS State rotate regularly and that review processes themselves are periodically revised, it is essential to establish stable communication channels among the partners. This would ensure a shared understanding of what each State reviews (scope), when reviews are conducted (timing), and how they are carried out (applicable legal standards). While much of this information is publicly available for Australia, the UK, and the US – and, in practice, the three have been engaging with one another on weapons reviews where necessary, making coordination comparatively straightforward – this may not be the case for some of the contemplated future AUKUS Pillar 2 participants. Should additional States join Pillar 2 – particularly those whose review mechanisms are less transparent or not well established, such as Japan or South Korea – ensuring mutual awareness of national legal review approaches and their differences becomes a critical starting point for effective coordination.
Second, the timing of reviews is another significant consideration. States may conduct legal reviews at different stages in the capability acquisition cycle and at varying speeds, with direct implications for jointly developed capabilities. For instance, where a State’s acquisition framework links the release of funding to the achievement of specific milestones – some of which may depend on a positive legal review – the overall pace of a joint programme may be determined by the slowest review process. Greater transparency among partners regarding such procedural requirements would facilitate better synchronisation of review timelines and more effective project planning.
Third is the issue of information and data sharing. The coordination of relevant information and data on the expected behaviour of the capability is critical for an effective legal review. Even within a single State, the conduct of meaningful legal reviews – particularly of complex, software-driven AI-enabled systems – depends on access to technical expertise and performance data that increasingly resides with industry rather than government.[89] In particular, legal review authorities would need to rely on structured collaboration with industry, especially for testing, evaluation, verification, and validation (TEVV) of capabilities. Yet such cooperation is frequently constrained by contractual limitations and proprietary concerns.[90]
These challenges risk being amplified in a trilateral setting, where the number of actors – governmental, military, and industrial – increases, and where sensitivities relating to intellectual property and national security are heightened. In a joint development context, different components of a capability may be designed and built by different partners, such that relevant data is not automatically accessible to all participating States. As some Workshop participants observed, this issue may arise, for example, where one State subcontracts the development of a component to a private contractor that imposes strict non-disclosure obligations vis-à-vis third parties. In such a scenario, there is a risk that the other AUKUS partners may be unable to access the data necessary to conduct their own legal reviews.
We believe that data sharing constraints can, to a significant extent, be mitigated within a TCM. In particular, participating States can identify, at an early stage, the categories of information required for their respective legal reviews and communicate these requirements to their partners. This would allow the necessary administrative processes for access and transfer to be initiated early in the development, reducing the risk of downstream delays. In addition, where relevant information is held by private contractors in another State, obligations to generate and share specified data can be embedded contractually, ensuring that required information is available for the purposes of national legal reviews in other partner States.
Fourth, coordination of (interim) review outcomes is even more critical, since these outcomes directly impact the development of a capability. In principle, reviews may yield three types of outcomes: positive (the capability is cleared for use), negative (the capability cannot be used lawfully and is thus not cleared for use), or conditional (the capability is cleared for use in certain specific circumstances and not cleared for use in certain other circumstances). In practice, outright negative findings are rare, while conditional outcomes are more common.
In a context of joint capability development, these outcomes have direct consequences for partner participation. An interim negative review by one AUKUS partner would likely require that partner to withdraw from the project, as it could not lawfully support or contribute to the capability. A conditional outcome may limit a State’s ability to support particular functions or uses of the system. Awareness of partners’ (interim) review outcomes is therefore essential. Where a State can no longer support certain aspects of a capability, the joint capability development must either be adjusted to meet its legal requirements or continue without that partner’s full participation. In the latter case, remaining partners would need to reassess the feasibility of the project, including funding arrangements and the redistribution of development and manufacturing responsibilities. A TCM would serve as a forum for the exchange of legal review outcomes, enabling their timely dissemination and facilitating coordinated responses among partner States. Importantly, this exchange need not involve the disclosure of full legal reviews. It may be limited to concise notifications identifying the aspect of the capability found to be unlawful or problematic and the resulting national position. Properly designed, such targeted information-sharing would avoid many of the sensitivities associated with any broader data disclosure.
Fifth is an issue of security clearances. The difficulty of aligning and recognising security clearance regimes across the three AUKUS partners may need to be addressed. Although there is broad structural similarity between national clearance and classification systems, there is no automatic mutual recognition, making the ‘translation’ of clearances a complex and often slow process.[91] This has direct implications for legal review: access to classified technical data, operational concepts, and performance assessments – often essential for meaningful analysis – depends on personnel holding the appropriate clearances within each jurisdiction. While national systems, such as Australia’s, are capable of allocating reviewers with the requisite clearance levels and managing sensitive or proprietary information through NDAs,[92] extending this model across three States introduces additional layers of administrative, legal, and security coordination. The result is a potential bottleneck in information access, personnel deployment, and timely review.
At the same time, as suggested by one of the Workshop participants who conducted legal reviews for Australia in the past, security clearance requirements are unlikely to constitute an insurmountable barrier in practice. Rather, the challenge lies less in the existence of clearance regimes as such, and more in ensuring that coordination and mutual recognition processes within a TCM operate in a timely and sufficiently robust manner to enable rapid and meaningful access to the information required for legal review.
Seventh, a TCM may be formal or informal. An informal mechanism, not grounded in a specific legal arrangement, may be easier to establish and more adaptable to the evolving needs of a particular Pillar 2 capability development program. However, given that one of the TCM’s core functions would be the exchange of often classified information – such as performance data relevant to legal reviews or the outcomes of those reviews – which typically requires formal intergovernmental arrangements, the AUKUS partners may ultimately need to consider establishing the mechanism on a formal basis.
Lastly, TMC should not be confused with transparency measures often discussed and called for in the context of legal reviews. Specifically, data and outcome sharing have been proposed in public discussions of legal reviews as such transparency measures.[93] These measures are primarily directed outward, aiming to build broader confidence in the conduct and integrity of legal reviews.
The purpose of the TCM, however, is not to promote transparency beyond the participating States, but to minimise friction in the joint development of capabilities under AUKUS Pillar 2 by facilitating timely and targeted information exchange among partners. In this sense, TCMs are instruments of internal coordination rather than outward-facing transparency measures.
Conclusion
During the drafting of AP I, several States proposed the creation of an international mechanism to assess the lawfulness of new weapons. That proposal failed to attract sufficient support.[94] Instead, Article 36 was framed in a manner that left legal reviews a distinctly national undertaking. This reflected a pragmatic recognition: unless States retained discretion to design review processes compatible with their own institutional and organisational structures and defence demands and practices, formalised and systematic legal review regimes might never emerge at all. The framework adopted in AP I thus embodies a conscious compromise – imposing a shared legal obligation while preserving decentralised, nationally tailored implementation.
The creation of the AUKUS partnership, and in particular Pillar 2 cooperation on advanced capabilities, effectively reopens this question in a narrower (at this stage) trilateral setting. Yet, as this paper argues, even among three closely aligned partners, a fully joint multinational legal review mechanism is neither realistic nor desirable. Persistent divergences in how reviews are scoped, as well as in the interpretation and application of relevant legal standards, present structural obstacles that seem unlikely to be overcome. In this context, a TCM emerges, in our view, as the only credible and operationally viable pathway. To be sure, it is not without challenges, including potential constraints on technical data-sharing, the need to develop and align TEVV standards, to ensure smoother mutual recognition of security clearances, and to establish or recognise trusted third-party evaluators. These are not structural barriers, however, but practical issues that can be addressed. A TCM would preserve national review processes while enabling targeted cooperation and structured information exchange. It would ultimately give effect to the spirit of Article 36 without demanding an unachievable degree of institutional alignment or convergence in the formulation and interpretation of legal standards.
Even if one accepts the view that, only a few years into the strategic partnership, both Pillars 1 and 2 are broadly where AUKUS leaders want them to be,[95] the time to think seriously about a trilateral legal review coordination mechanism is now. If such a mechanism is to support Pillar 2 effectively, the partners must already be considering how to design it and bring it into being.
Endnotes
[1] AUKUS is a trilateral security partnership announced in September 2021 between Australia, the United Kingdom and the United States structured around two principle lines of effort: Pillar I concerns Australia’s acquisition of conventionally armed, nuclear-powered submarines; Pillar II concerns cooperation on advanced defence capabilities, including cyber, AI, quantum, undersea systems, hypersonics and electronic warfare. See Luke A. Nicastro, “AUKUS Pillar 2 (Advanced Capabilities): Background and Issues for Congress” (Congressional Research Service, May 2024) 4 <https://sgp.fas.org/crs/row/R47599.pdf>.
[2] The White House, “Joint Leaders Statement on AUKUS,” September 15, 2021, https://bidenwhitehouse.archives.gov/briefing-room/statements-releases/….
[3] William Greenwalt and Tom Corben, “AUKUS Enablers? Assessing Defence Trade Control Reforms in Australia and the United States” (Sydney: United States Studies Centre at the University of Sydney, August 2024), 8, https://cdn.sanity.io/files/ooh1fq7e/production/152c48a93f6ac0a7bc04719….
[4] Amy McDonnell, “Uncertified: The Implications of the US Department of State’s Decision on Australia’s ITAR Exemption” Export Controls Australia Group, 1 May, 2024, https://www.ecag.com.au/post/uncertified-the-implications-of-the-us-dep….
[5] Nicastro, “AUKUS Pillar 2”, 4.
[6] McFarland, T., Assaad, Z., “Legal reviews of in situ learning in autonomous weapons”. Ethics Inf Technol 25, 11 (2023). Sanders, L., & Copeland, D. (2020, November 27). Developing an Approach to the Legal Review of Autonomous Weapon Systems. ILA Reporter. https://ilareporter.org.au/2020/11/developing-an approach-to-the-legal-review-of-autonomous-weapon-systems lauren-sanders-and-damian-copeland/; Klaudia Klonowska, "Article 36: Review of AI Decision-Support Systems and Other Emerging Technologies of Warfare," Asser Research Paper No. 2021-02 (The Hague: T.M.C. Asser Institute for International & European Law, March 17, 2021), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3823881.
[7] Austin Wyatt et al, “Towards AUKUS Collaboration on Responsible Military Artificial Intelligence”, RAND Australia, 2024 18, www.rand.org/content/dam/rand/pubs/research_reports/RRA3000/RRA3079-1/R….
[8] See section C below.
[9] James Palmer, “AUKUS needs a third pillar: space”, Australian Strategic Policy Institute, 27 June, 2025, https://www.aspistrategist.org.au/aukus-needs-a-third-pillar-space/.
[10] Nicastro, “AUKUS Pillar 2”, 1.
[11] The White House, “Joint Leaders Statement”.
[12] United Kingdom Government, “Fact Sheet: Implementation of the Australia – United Kingdom – United States Partnership (AUKUS)”, 5 April, 2022, https://www.gov.uk/government/publications/implementation-of-the-austra….
[13] Nicastro, “AUKUS Pillar 2”, 1.
[14] Michael Shoebridge, “What is AUKUS and what is it not?”, 2021 ASPI Strategist, 4, https://ad-aspi.s3.ap-southeast-2.amazonaws.com/2021-12/What%20is%20AUK….
[15] Australian Strategic Policy Institute, “AUKUS Relevant Technologies: Top 10 Country Snapshot”, 2023 1 https://ad-aspi.s3.ap-southeast-2.amazonaws.com/2023-06/PB69-CriticalTe…;
[16] Ibid.
[17] Central Committee of the Communist Party of China (CPC) and People’s Republic of Chine State Council, “Outline of the National Innovation-Driven Development Strategy Issued by the CPC Central Committee and the State Council,” trans. Center for Security and Emerging Technology, 11 December, 2019, https://cset.georgetown.edu/publication/outline-of-the-national-innovat….
[18] See also Arthur R. Kroeber, “Unleashing “New Quality Productive Forces”: China’s Strategy for Technology-led Growth”, The Brookings Institution, 4 June, 2024, https://www.brookings.edu/articles/unleashing-new-quality-productive-fo….
[19] Sarah Tzinieris et al, “Shedding Light on Chinese Thinking on AUKUS”, Pacific Focus Vol. XXXIX, No. 3 (2024): 504, https://doi.org/10.1111/pafo.12262.
[20] Kim Beazley, “We need a third pillar of AUKUS: critical minerals”, Australian Strategic Policy Institute, 23 May, 2024, https://www.aspistrategist.org.au/we-need-a-third-pillar-of-aukus-criti….
[21] Adam Thorn, “To infinity and beyond! Former PM calls for AUKUS to expand to space”, Defence Connect, 28 May 2025, https://www.defenceconnect.com.au/joint-capabilities/16125-to-infinity-…; See also Palmer, “AUKUS needs a third pillar”.
[22] See Natalia Jevglevskaja, “AUKUS Pillar 2: Legal Implications of Sharing Advanced Technology” [pp xy], in Armed Conflict and International Law in the Indo-Pacific Region, Douglas Guilfoyle and Jennifer Maddocks, Oxford University Press, 2026, doi: 10.1093/oso/9780197844540.003.0014. See also Lisa Curtis and Ryan Claffey, “Congressional support key for sustaining momentum of US, UK, Australia partnership”, The Hill, 02 February, 2026, https://thehill.com/opinion/national-security/5717755-aukus-defense-ind….
[23] Tom Corben and Sophie Mayo, “Patience, please. AUKUS Pillar 2 is indeed making progress”, Australian Strategic Policy Institute, 19 April, 2024, https://www.aspistrategist.org.au/patience-please-aukus-pillar-2-is-ind…; Peter Tesch, “AUKUS: Two pillars, three fallacies”, The Lowy Institute, 13 March, 2025, https://www.lowyinstitute.org/the-interpreter/aukus-two-pillars-three-f…; Molly Donohue Magee, “AUKUS Tech Development A Marathon, Not Sprint”, National Defense Magazine, 18 July, 2024, https://www.nationaldefensemagazine.org/articles/2024/7/18/aukus-tech-d….
[24] American Chamber of Commerce in Australia and Business Council of Australia partnership. “Australia’s AUKUS Pillar II Opportunity”, November, 2025, 5, https://www.bca.com.au/wp-content/uploads/2025/11/Australias-AUKUS-Pill….
[25] Peter Dean and Alice Nason, “AUKUS Pillar II Is Failing in Its Mission”, War on the Rocks, 2 June, 2025, https://warontherocks.com/2025/06/aukus-pillar-ii-is-failing-in-its-mission-it-needs-its-own-optimal-pathway/#:~:text=Pillar%20I%20submarine%20cooperation%20remains,Class%20Justin%20Yarborough%20via%20DVIDS;
George Allision, "AUKUS Pillar Two ‘lacks focus and delivery’ say MPs", UK Defence Journal, 14 December 2025, https://ukdefencejournal.org.uk/aukus-pillar-two-lacks-focus-and-delivery-say-mps/.
[26] Nicola Macaulay, “Is NZ defence and intelligence policy aligning with AUKUS in all but name?”, The Conversation, February 4, 2026, https://theconversation.com/is-nz-defence-and-intelligence-policy-align…; Adam Broinowski, “AUKUS Pillar 2, Japan and South Korea”, Parliamentary Library, August 15, 2024, https://parlinfo.aph.gov.au/parlInfo/download/library/prspub/9893294/up…; Alexander Gray and Doug Stokes, “From AUKUS to CAUKUS: The Case for Canadian Integration”, Legatum Institute, February 2024, https://www.prosperity.com/wp-content/uploads/2024/02/AUKUS-PILLAR-2.pdf.
[27] See also Simon McKenzie and Eve Massingham, “AUKUS: The Regulation of the Ocean and the Legal Dangers of Working Together” Ocean Yearbook Online 37, No 1 (2023) 148-152.
[28] Protocol Additional to the Geneva Conventions of 12 August 1949, and Relating to the Protection of Victims of International Armed Conflicts (Protocol I), opened for signature 8 June 1977, 1125 UNTS 3 (entered into force 7 December 1978).
[29] The reference to assessment “prior” to development or acquisition should not be understood as implying that the review obligation is exhausted once a capability is fielded. Rather, it identifies the minimum point at which legal scrutiny must occur. Where circumstances so require – such as where a system is materially modified, thereby producing effects different from those originally reviewed, or is deployed in novel ways – further review may be necessary after deployment. See Netta Goussac et al., “Enhancing the Legal Review of Autonomous Weapon Systems: Report of an Expert Meeting (Sydney, 28–30 March 2023)”, Law and the Future of War Research Group, TC Beirne School of Law, The University of Queensland, 1 May, 2023, 11, https://doi.org/10.14264/2bbfd31
[30] Natalia Jevglevskaja, International Law and Weapons Review: Emerging Military Technology under the Law of Armed Conflict (Cambridge University Press, 2021) ch 5.5.
[31] See Goussac et al., “Enhancing the Legal Review”, sec 2.6; ‘The review examined the Winchester pump action shotgun firing buckshot pellets to be employed in compliance with the customary law rule prohibiting the use of particularly injurious weapons (now codified in Art. 35(2) AP I)’, see Renato Wolf et al., “Advancing the Legal Review of Autonomous Weapon Systems: Report of an Expert Meeting (Sydney, 16–18 April 2024))”, Law and the Future of War Research Group, TC Beirne School of Law, The University of Queensland, 1 September, 2024, 12, https://doi.org/10.14264/c41c6fb.
[32] See Wolf et al. “Advancing the Legal Review”, 12.
[33] Australian Department of Defence, “Defence Instruction (General) OPS 44-1: Legal Review of New Weapons”, 2 June, 2005.
[34] Australian Department of Defence, “Defence Article 36 Reviews of New Weapons Guide”, May, 2020.
[35] Department of Defence, “Australia’s Guide to the Legal Review of New Weapons, Means or Methods of Warfare”, 10 April, 2024, https://legalreviewportal.org/files/au/2024-04-10-adf_guide_to_art_36_l….
[36] Department of Defence, “AF172: Article 36 Legal Review Report”, 16 November, 2023.
[37] Wolf et al. “Advancing the Legal Review”, 11.
[38] Damian Copeland and Rain Liivoja, “Progressing the Legal Review of Autonomous Weapon Systems: Report of an Expert Meeting (Geneva, 10-11 March 2025)”, The Asia-Pacific Institute for Law and Security, September, 2025, 9, https://apils.org/2025/09/04/report-of-the-2025-expert-meeting-on-the-l….
[39] Copeland and Liivoja, “Progressing the Legal Review” 9. See also United Kingdom Ministry of Defence Development, Concepts and Doctrine Centre, “UK Weapon Reviews” (2016), https://assets.publishing.service.gov.uk/media/5a80bf5f40f0b62305b8cec5….
[40] Copeland and Liivoja, “Progressing the Legal Review”, 9.
[41] Wolf et al. “Advancing the Legal Review”, 11.
[42] Copeland and Liivoja, “Progressing the Legal Review”, 10.
[43] Wolf et al. “Advancing the Legal Review”, 12.
[44] Copeland and Liivoja, “Progressing the Legal Review”, 10.
[45] Copeland and Liivoja, “Progressing the Legal Review”, 10.
[46] A collaborative organisation that brings together industry, academia, government and defence stakeholders across the three AUKUS countries to promote cooperation, innovation and economic opportunities arising from the partnership, see “AUKUS Forum”, accessed 20 March, 2026, https://aukusforum.com/.
[47] Allyson Park, “U.S., U.K., Australia Still Ironing Out Tech, Info Sharing Kinks”, National Defense Magazine, 7 March, 2025, https://www.nationaldefensemagazine.org/articles/2025/3/7/us-uk-austral….
[48] Note that in the US, certain technology, such as AWS, is subject to policy-level assessment at the Department of Defense level in addition to a legal review of the weapon system conducted by the respective service. See Department of Defense. “DoD Directive 5000.01, The Defense Acquisition System” United States of America, July 28, 2022 https://www.esd.whs.mil/Portals/54/Documents/DD/issuances/dodd/500001p…; Department of Defense. “Directive Number 3000.09: Autonomy in Weapon Systems.” United States of America, May 8, 2017. https://www.esd.whs.mil/portals/54/documents/dd/issuances/dodd/300009p…; Wolf et al. “Advancing the Legal Review” 12-13).
[49] Vincent Boulanin and Maaike Verbruggen, “Article 36 Reviews: Dealing with the Challenges Posed by Emerging Technologies,” Stockholm International Peace Research Institute, December, 2017, 15, https://www.sipri.org/sites/default/files/2017-12/sipri_bp_1712_article….
[50] See also Netta Goussac and Rain Liivoja, “Legal Reviews of Military Artificial Intelligence Capabilities”, Asia-Pacific Institute for Law and Security, 25 August, 2025, https://apils.org/2025/08/25/legal-reviews-of-military-artificial-intel….
[51] However, see also our comment in FN 28 above.
[52] Group of Governmental Experts on Emerging Technologies in the Area of Lethal Autonomous Weapons Systems “Australia’s System of Control and Applications for Autonomous Weapon Systems”, 26 March, 2019, https://undocs.org/CCW/GGE.1/2019/WP.2/Rev.1.
[53] Ibid, para 3.
[54] Goussac et al., “Enhancing the Legal Review”, 11.
[55] DoD Directive 2311.01: DoD Law of War Program (2 July 2020) s 1.4. See also Andrew McKee and Christina Colclough, “Legal Reviews of Autonomous Weapons at the Tactical Edge,” Articles of War, Lieber Institute West Point, November 12, 2025, https://lieber.westpoint.edu/legal-reviews-autonomous-weapons-tactical-edge/.
[56] This understanding draws on in-person discussions with officials responsible for legal reviews in Australia, the UK and the US, held during the Sydney Expert Meetings on Autonomous Weapon Systems. See generally Goussac et al., “Enhancing the Legal Review” and Wolf et al. “Advancing the Legal Review”.
[57] Department of Defence, “Australia’s Guide to the Legal Review”, sec. 5(a)–(b).
[58] United Kingdom Ministry of Defence Development, Concepts and Doctrine Centre, “UK Weapon Reviews”, 4-5. (While the current Joint Services Publication is classified, there are no indications in the publicly available material that the UK has changed its treatment of ‘weapons / means’).
[59] United States, “Department of the Air Force Instruction 51-401: The Law of War,” 3 August, 2018. pt 2, para. 5 and Glossary, https://static.e-publishing.af.mil/production/1/af_ja/publication/afi51…; United States, “Department of the Army Regulation 27-53: Legal Review of Weapons and Weapon Systems”, 23 September, 2019, Glossary; United States, “Secretary of the Navy Instruction 5000.2F: Defense Acquisition System and Joint Capabilities Integration and Development System Implementation”, 26 March, 2019, encl 3, s 10(e)(1).
[60] United States, “Department of the Air Force Instruction 51-401”, para. 5 and Glossary (emphasis added).
[61] United States, “Department of the Air Force Instruction 51-401”, pt 2, para. 5 and Glossary; United States, “Department of the Army Regulation 27-53”, sec 5(d)(2); United States, “Secretary of the Navy Instruction 5000.2F”, encl 3, s 10(e)(3).
[62] Department of Defence, “Australia’s Guide to the Legal Review”, Part I, s 5(c).
[63] Wolf et al. “Advancing the Legal Review”, 11.
[64] United States, “Department of the Air Force Instruction 51-401”, pt 2, sec 6.1.2; United States, “Department of the Army Regulation 27-53”, sec 8(b); United States, “Secretary of the Navy Instruction 5000.2F”, encl 3, s 10(f)(2).
[65] This understanding is informed by in-person discussions with officials responsible for legal reviews in the United States, held during the Sydney Expert Meetings on Autonomous Weapon Systems. See generally Goussac et al., “Enhancing the Legal Review” and Wolf et al. “Advancing the Legal Review”.
An example is Department of Defense Directive 3000.09 on Autonomy in Weapon Systems, which requires that such systems ‘be designed to allow commanders and operators to exercise appropriate levels of human judgment over the use of force’, and that the system design incorporates the capabilities necessary to support this requirement before development proceeds. See Department of Defense, ‘Directive Number 3000.09: Autonomy in Weapon Systems’ https://www.esd.whs.mil/portals/54/documents/dd/issuances/dodd/300009p…, Art. 4(1)(c)(1).
[66] Art. 35(2) AP I and International Committee of the Red Cross Customary IHL Database, “Rule 70. Weapons of a Nature to Cause Superfluous Injury or Unnecessary Suffering,” accessed May 15, 2026.
[67] Legality of the Threat or Use of Nuclear Weapons (Advisory Opinion) [1996] ICJ Reports 226, p. 35, para. 78.
[68] Rain Liivoja, “Protecting Warfighters from Superfluous Injury and Unnecessary Suffering”, in “Civility, Barbarism and the Evolution of International Humanitarian Law”, ed. Matt Killingsworth and Tim McCormack (University of Oxford Press 2024), 189.
[69] US Department of the Army (1956). Field Manual (FM) 27-10: The Law of Land Warfare, Washington, DC: Department of the Army.
[70] W. Hays Parks, “Conventional Weapons and Weapons Reviews”. 8 Yearbook of International Humanitarian Law 55, 86-87 footnote 123 (2005); Richard R Baxter, “Conventional Weapons Under Legal Prohibitions”, 1 International Security 42, 43 (1977).
[71] U.S. Department of Defense, Department of Defense Law of War Manual (Washington, DC: Office of General Counsel, 2015; updated July 2023), section 6.6.
[72] US Law of War Manual, section 6.6.1. For an explanation as to why the US is in the minority on this, see ibid, Rain Liivoja, “Protecting Warfighters from Superfluous Injury and Unnecessary Suffering”, 188-189.
[73] Cf Liivoja, “Protecting Warfighters from Superfluous Injury and Unnecessary Suffering”, 188.
[74] Jean-Marie Henckaerts and Louise Doswald-Beck, Customary International Humanitarian Law—Volume I: Rules, (Cambridge University Press 2005), Rule 45; See also Saeed Bagheri and Gerhard Kemp, “Damage to and Destruction of the Natural Environment: Terraforming Warfare in Gaza and Accountability for Ecocentric Crimes”, Journal of International Criminal Justice 23, no 3-4 (2025): 436, https://doi.org/10.1093/jicj/mqaf049.
[75] United States, Departement of Defense Law of War Manual, (Office of General Counsel, July 2023), sec. 6.10.3.1.
[76] United States, “Department of the Air Force Instruction 51-401”, pt 2, para. 5; United States, “Department of the Army Regulation 27-53”, sec. 6(a); United States, ’Secretary of the Navy Instruction 5000.2F”, encl. 3, s 10(a).
[77] Department of Defence, “Australia’s Guide to the Legal Review”, part 1 section 2 and part 2 sections 10, 19-21 and 24.
[78] Copeland and Liivoja, “Progressing the Legal Review”, 9-10.
[79] This follows from the foundational principle of international law – the sovereign equality of States, Article 2(1) of the UN Charter, Charter of the United Nations, June 26, 1945, 1 U.N.T.S. XVI. See also Treatment of Polish Nationals and Other Persons of Polish Origin or Speech in the Danzig Territory, Advisory Opinion, 1932, P.C.I.J., Series A/B, No. 44, pp. 24-25. See also Lotus, Judgment No. 9, 1927, P.C.I.J., Series A, No. 10, p. 24. See also International Law Commission, Draft articles on Responsibility of States for Internationally Wrongful Acts, with commentaries, UN Doc A/56/10, Article 3, para. 2 https://legal.un.org/ilc/texts/instruments/english/commentaries/9_6_200….
[80] Convention on Cluster Munitions, opened for signature 3 December 2008, 2688 UNTS 39 (entered into force 1 August 2010).
[81] Natalia Jevglevskaja, International Law and Weapons Review: Emerging Military Technology under the Law of Armed Conflict, (Cambridge University Press 2021), ch. 4.1.
[82] United States, “Department of the Army Regulation 27-53”, glossary and sec. 8.
[83] United States, “Department of the Air Force Instruction 51-401”, Attachment 1 and part 2 section 7.
[84] Peter Dean and Alice Nason, “AUKUS Pillar II Is Failing in Its Mission”, War on the Rocks, 2 June, 2025, https://warontherocks.com/2025/06/aukus-pillar-ii-is-failing-in-its-mis….
[85] Ibid.
[86] See section C(2)(ii) above discussing the standard of review.
[87] The criticisms focus in particular on the substance and operation of the International Traffic in Arms Regulations (ITAR) administered by the Directorate of Defence Trade Controls (DDTC) [ITAR, 22 CFR §§120-130]. The prominent ‘eight deadly sins’ if ITAR include ‘universality and non-materiality’, ‘extraterritoriality’, ‘non-discrimination between different classes of recipients’, ‘transactional process-compliance’, ‘knowledge taint’; ‘non-reciprocity’, and ‘unwarranted predictability’. On their detailed substantiation, see William Greenwalt and Tom Corben, “Breaking the Barriers: Reforming US Export Controls to Realise the Potential of AUKUS”, United States Studies Center, 2023, 10-16, https://www.ussc.edu.au/breaking-the-barriers-reforming-us-export-contr….
[88] S 1343 NDAA (amending Section 38 of the Arms Export Control Act (22 U.S.C. 2778)). This certification was issued to Congress on 15 August 2024, see The US Department of Defence, “AUKUS Defense Trade Integration Determination”, 15 August, 2024, https://2021-2025.state.gov/aukus-defense-trade-integration-determinati….
[89] Goussac and Liivoja, “Legal Reviews of Military Artificial Intelligence Capabilities”.
[90] Daniel Fiott, “Military-technological cooperation across the Euro-Atlantic and Indo-Pacific”, War on the Rocks, 19 January, 2024, https://warontherocks.com/2024/01/military-technological-cooperation-across-the-euro-atlantic-and-indo-pacific/.
[91] American Chamber of Commerce in Australia and Business Council of Australia partnership. “Australia’s AUKUS Pillar II Opportunity”, November, 2025, 5, https://www.bca.com.au/wp-content/uploads/2025/11/Australias-AUKUS-Pill….
[92] See Department of Defence, “Australia’s Guide to the Legal Review”, Part I, sections 21-22.
[93] Wolf et al., “Advancing the legal review”, p. 19.
[94] See Natalia Jevglevskaja, International Law and Weapons Review: Emerging Military Technology under the Law of Armed Conflict (Cambridge University Press, 2021) ch 5.1. See also Netta Goussac and Rain Liivoja, “Legal Reviews of Military Artificial Intelligence Capabilities“, Asia-Pacific Institute for Law and Security, 25 August, 2025, https://apils.org/2025/08/25/legal-reviews-of-military-artificial-intelligence-capabilities.
[95] Kris Osborn, “AUKUS Nations Shift Priorities From R&D to Deliverable Technologies,” National Defense Magazine, June 6, 2025, https://www.nationaldefensemagazine.org/articles/2025/6/6/aukus-nations….