Ukrainian President Volodymyr Zelensky (C) and NATO Secretary General Jens Stoltenberg speak to the media on the second day of the 2023 NATO Summit on July 12, 2023 in Vilnius, Lithuania. The summit is bringing together NATO members and partner countries heads of state from July 11-12 to chart the alliance's future, with Sweden's application for membership and Russia's ongoing war in Ukraine as major topics on the summit agenda. (Photo by Sean Gallup/Getty Images)
Source: Getty

Getting Ukraine’s Security Agreements Right

The security agreements between Ukraine and its partners contain important commitments for ensuring Ukraine’s long-term defense and deterrence capabilities. Still, there is significant room to make their design more credible and durable.

by Mykhailo Soldatenko
Published on July 8, 2024

In July 2023, on the margins of the NATO summit in Vilnius, the leaders of the G7 countries issued a joint declaration pledging to work with Ukraine on long-term bilateral security commitments with the aim of bolstering its self-defense and deterrence capabilities until it becomes a member of NATO. More than two dozen other nations subsequently endorsed the declaration and pledged to negotiate their own bilateral security agreements with Ukraine. Under the Vilnius framework, Ukraine has signed ten-year security agreements with twenty partners, including all G7 members (Canada, France, Germany, Italy, Japan,1 the United Kingdom, and the United States) and Belgium, Denmark, Estonia, Finland, Iceland, Latvia, Lithuania, the Netherlands, Norway, Poland, Portugal, Spain, and Sweden. Ukraine also signed a separate agreement with the European Union.2

Ukrainian President Volodymyr Zelensky has touted these agreements as the “new pillars for the rules-based international order.” But some commentators have compared them to the ambiguous Budapest Memorandum, in which the United States, the United Kingdom, and Russia provided Ukraine with relatively weak security commitments in 1994 in exchange for its denuclearization. While the new agreements fall short of mutual defense obligations (colloquially known as “security guarantees”)—they do not oblige the signatories to send their armed forces to Ukraine’s defense—they nonetheless contain more meaningful and robust commitments than those in the Budapest Memorandum.

In the new agreements, Ukraine’s partners committed, at a minimum, to maintaining the type and scale of military and other support they have provided since the start of Russia’s full-scale invasion in 2022. That alone is a significant development for Kyiv, which had no such commitments on paper before then. The partners also pledged to enable Ukraine to deter and, if necessary, defend itself from future Russian armed attacks after current hostilities end. To do so, they vowed to build Ukraine’s “future force” and to consult with Kyiv and take further measures if the country comes under armed attack again in the future.

That said, there is significant room to make the commitments more concrete, resilient, and durable in order to credibly signal to Russia that Ukraine will receive the support it needs. The partners and Ukraine should consider:

  1. Outlining more detailed goals for their military aid to Ukraine based on a unified vision and common set of measurable standards;
  2. Projecting cost estimates of support for the entire ten-year duration of the agreements;
  3. Institutionalizing and formalizing the agreements into a coherent framework; and
  4. Strengthening the legal basis of the agreements, including by securing cross-party buy-in from national legislatures.

The upcoming NATO summit in Washington may be a good occasion for the signatories to lay out a common vision for implementing the Vilnius G7 declaration, including by potentially signing a multilateral instrument connecting all the security agreements in a single framework.

A Mixed Bag of Goals for Military Support to Ukraine

In order to design and resource Ukraine’s future force, Kyiv and its partners must agree on a common vision and set of standards for military assistance. But the various security agreements invoke a patchwork of terminology that points to different goals and levels of commitment among the signatories.

Bolstering Ukraine’s defense and deterrence capabilities is a common theme that runs throughout the agreements. Most of them codify, with variations in the wording, the Vilnius declaration’s aim of “ensuring a sustainable force capable of defending Ukraine now and deterring Russian aggression in the future” (emphasis added). The U.S. text elevates the goal of assisting Ukraine to maintain a “credible defense and deterrence capability” as a central organizing principle.

But not all of the agreements link further military assistance with Ukraine’s goal of liberating its entire territory. This discrepancy is not merely a symbolic one: territorial liberation is a more ambitious military objective than defense and deterrence. The former requires more advanced offensive capabilities and greater proficiency in combined arms–maneuver warfare. But the fact that some of Ukraine’s partners hedged suggests that it might be enough for them if Kyiv could simply defend the territory it currently controls and deter further Russian advances in the future. Ukraine’s public signaling and the stronger wording it agreed on with certain partners suggest there is no consensus on the goals of military support. 

The French, Italian, Lithuanian, Portuguese, and UK agreements are the strongest in this respect. In addition to the defense and deterrence standard, they commit the signatories to work together “to ensure [the Ukrainian defense and security forces] are able to fully restore Ukraine’s territorial integrity within its internationally recognised borders.”3  Many other agreements invoke the restoration of Ukraine’s territorial integrity as a goal of military support, but they do not explicitly link this goal with commitments to ensure sufficient Ukrainian capabilities.4

Although the U.S.-Ukraine agreement mentions restoring Ukraine’s territorial integrity as an overarching policy goal in the section on a “just peace,” it does not apply the restoration standard as a goal of military support. The agreement also stipulates the U.S. intention “to support Ukraine’s efforts to win today’s war.” But the meaning of victory is not defined.

Several of the agreements contain a benchmark for support that resembles the U.S. commitment to maintain Israel’s “qualitative military edge” (QME). Belgium, Italy, the Netherlands, and Poland declare their policy to help Ukraine with preserving its “qualitative defence and military edge.” (France uses the term “qualitative defence and military capabilities.”) Denmark, Norway, and Poland mention Ukraine’s “military edge” in the context of defense industrial cooperation. Other agreements, including the U.S. one, do not contain similar wording.

Belgium, Canada, Estonia, France, Latvia, Lithuania, the Netherlands, Poland, and the United Kingdom—with some differences in wording—also commit to seeking to ensure that Ukraine’s military capabilities are “at such a level that, in the event of external military aggression against [the relevant partner], Ukraine is able to provide effective military assistance” to it. This standard highlights Ukraine’s potential contributions to NATO and value as a security provider.5 Of course, to provide such aid, Ukraine would first have to be able to meet its own security needs. But the standard may still be relevant when evaluating Ukraine’s own deterrence capability vis-à-vis Russia.

Ukraine and its partners should consider standardizing the terminology for these goals and benchmarks and creating agreed definitions against which progress can be measured. Phrases like “credible defense and deterrence capability” and “[the ability to] fully restore Ukraine’s territorial integrity” are useful to frame long-term assistance. But they are too vague to mean much when it comes to evaluating Ukraine’s actual military strength and calibrating aid accordingly.

Take the example of the U.S. commitment to Israel’s QME, which the U.S. Congress defined in law in 2008.6 The legal definition of QME is specific enough to allow Israeli and American policymakers to judge whether the standard is being met and what adjustments need to be made. Moreover, U.S. law requires that the president “carry out an empirical and qualitative assessment on an ongoing basis of the extent to which Israel possesses a qualitative military edge over military threats to Israel.” If Ukraine and its partners can agree on common terms and definitions, they will be in a better position to carry out similar assessments of Ukraine’s capabilities relative to those of Russia and adjust future support accordingly.

Partners Come Up Short on Funding

To achieve these ends, the agreements detail a range of short- and long-term types of support, building on the initial framework outlined in the Vilnius G7 declaration. Most of the agreements specify priority areas for weapons deliveries, to include air defense, artillery, long-range fires, armored vehicles, and air power. But the commitments are quite general and do not mention specific systems or delivery schedules.

The nonbinding annex to the U.S.-Ukraine agreement specifies, in greater detail than the other agreements, the domains of intended (rather than committed) military support. Within each domain, the United States made relatively specific political commitments. For example, the United States committed to supporting Ukraine’s development of “a joint fires capability” and “layered, integrated air and missile defense system,” and to working with other allies “on the modernization of Ukraine’s Air Force.” This level of detail on specific capabilities goals is a relative advantage of the U.S. agreement over others. 

The agreements also mention other forms of security cooperation, such as support for the development of Ukraine’s indigenous defense industry, training and exercises, intelligence sharing, and countering cyber and informational threats. That said, the concrete commitments in these areas tend to be thin. One exception is the U.S.-Ukraine agreement, which specifies that the parties are obligated to “advance the appropriate sharing of intelligence and to promote enhanced cooperation between their intelligence services, with the scope and procedure of cooperation determined by their respective entities responsible for intelligence and security.” Presumably those details will be worked out in classified implementing agreements.

When it comes to concrete numbers, the agreements underwhelm. It is challenging to project how much it will cost to build and maintain Ukraine’s capabilities over ten years amid active hostilities and without clarity about the war’s end. Still, most partners stipulate the level of military assistance only for 2024, with no promises in future years. Notably, the Danish, Norwegian, and Swedish agreements envisage multiyear support programs that can set an example for other partners. Similarly, Estonia, Latvia, and Lithuania define multiyear spending targets of 0.25 percent of their respective annual gross domestic product on security and defense support to Ukraine. Firm multiyear commitments would improve the predictability of aid so that Ukraine can better plan and budget for its future force. They would also send a more credible signal to the Kremlin.

The key to long-term planning and resourcing will be the U.S. contribution, given its relative size and the types of capabilities and leadership that only Washington can provide. The U.S.-Ukraine agreement states that the United States intends to seek Congress’s approval for future appropriations “to help sustain a Ukrainian credible defense and deterrent capability, in war and peace.” This is different from the funding model in the ten-year U.S.-Israel memorandum of understanding, which specified a fixed annual dollar amount that the executive branch committed itself to requesting from Congress. In Ukraine’s case, the White House may have been wary of specifying a dollar amount after having endured a bruising battle for passage of the $61 billion supplemental aid package. That said, the agreement creates a policy framework under which the U.S. president and Congress might be able to work toward specific figures, which would significantly increase the credibility and deterrent value of the commitments.

Improved Coordination Mechanisms

The agreements begin to codify a multinational framework for coordination among Ukraine’s partners through the capability coalitions within the Ukraine Defense Contact Group (UDCG, colloquially known as the Ramstein Group). As detailed in table 1, most of the agreements commit Ukraine’s partners to leading or participating in individual coalitions (or note their intention to do so). These working groups are tasked with developing plans that will guide Kyiv’s acquisition and training decisions, with the goal of rationalizing and modernizing Ukraine’s armed forces.

The UDCG and capability coalitions are a good start toward building an institutional framework for coordinating partner support and for developing standards and methods of evaluation. Further institutionalizing and formalizing the UDCG would help connect the patchwork of security agreements into a coherent multilateral policy framework, which could, in turn, increase the overall credibility of the commitments and make them less susceptible to elections or other political shocks. This process appears to be ongoing: the United Kingdom and Germany state in their respective agreements that they are working “to develop the governance framework for the Capability Coalitions.” (The UK agreement specifies that it is doing so alongside the United States, but the U.S. agreement does not mention it.) 

Responding to Future Attacks

Each agreement includes a mechanism to trigger consultations and agreed responses, including increased military support if Russia were to attack Ukraine again in the future.7 With some variations in wording, all of the agreements stipulate that, upon the request of either of the participants, the signatories will consult within twenty-four hours to determine “measures needed to counter or deter the aggression” (as in the agreements with France, Italy, and the United Kingdom) or, less concretely, “appropriate next steps” (as in the agreements with the Netherlands and the United States8). Most agreements, except the U.S. one, also include a version of the following committed9 responses:

  • To provide Ukraine with swift and sustained security assistance, modern military equipment across all domains as necessary, and economic assistance;
  • To impose economic and other costs on Russia; and
  • To consult with Ukraine on its needs as it exercises its right of self-defense enshrined in Article 51 of the United Nations (UN) Charter.

The U.S. agreement differs from the others in that it includes a legally binding commitment to meet, within twenty-four hours where possible, “at the most senior levels to determine appropriate next steps and additional defense needs.” It also connects the mechanism to threats of armed attack, not only to attacks. But the specific steps outlined in the main text are not binding, as they are embedded in sentences that include the verb “may.” The nonbinding annex to the agreement adds that the United States “intends” (not commits) to coordinate with Ukraine and other partners “on the potential need to rapidly increase the scope or scale of United States security assistance to Ukraine, including potentially the provision of additional weapon systems and equipment, and other materiel, as well as the exchange of information with Ukraine.”  

Importantly, the trigger for the response mechanism in each agreement is an armed attack (a Russian attack in most agreements but any armed attack in the U.S. one), which is the legal term of art from Article 51 on the right of self-defense in the UN Charter. The same trigger activates, for example, the collective self-defense provision in Article 5 of the North Atlantic Treaty. In international law, an armed attack is defined as the gravest form of unlawful use of force of sufficient scale and effect. Mere frontier incidents, slight provocations, cyber attacks without effects comparable to conventional armed attacks, and other insignificant uses of force would generally not be considered armed attacks and thus would not trigger the mechanism. That said, the United States is an outlier in this respect as it maintains the “no gap” theory,10 according to which any use of force constitutes an armed attack within the meaning of Article 51 of the UN Charter. This might lower the bar for triggering the mechanism in the U.S.-Ukraine agreement.  

The Canadian agreement clarifies that the trigger mechanism applies “following the cessation of current hostilities,” while the Lithuanian and Polish ones specify that it would apply “in the event of significant escalation of the ongoing aggression.” Other agreements merely use the qualifier “future” to describe the response mechanism’s trigger, which could potentially allow it to be activated if, for example, Russia launched another major offensive within the context of the current war.

In addition, all of the agreements except the U.S. one note that the partners and Ukraine may amend the mechanism to align it with any mechanism Ukraine agrees on with other partners. This is a clear acknowledgment that the mechanism is a work in progress, and that the desired end state is a collective response rather than a set of individual responses. Kyiv and its partners should seek to harmonize the wording of their respective response mechanisms and, considering the crucial nature of future U.S. support for its success, strengthen that provision in the U.S.-Ukraine agreement.

Legal Ambiguities Abound

Ukraine’s initial proposal, known as the Kyiv Security Compact, envisaged that the country’s partners would conclude legally binding security agreements. However, only the U.S.-Ukraine agreement has significant features of a legally binding treaty, including the use of treaty-like terms such as “parties” (instead of “participants” in other agreements),11 “shall” (in some parts), and “entry into force.”

U.S. President Joe Biden appears to have signed the document on his sole presidential authority as an executive agreement, which does not require congressional involvement. That said, the legal form does not mean that all of an agreement’s terms are legally binding. Indeed, much of the U.S.-Ukraine agreement sets forth U.S. “intentions” or “policies,” not legal obligations. Moreover, many key provisions are set forth in the explicitly nonbinding annex.12

There are also reasonable doubts as to whether Zelensky had the power to conclude such a consequential legally binding agreement with the United States without submitting it to the Verkhovna Rada (Ukraine’s legislature) for ratification.13 Among other things, the lack of parliamentary ratification creates uncertainty about the provisions of the U.S.-Ukraine agreement that purport to place a legal obligation upon Ukraine to “undertake efforts” toward the implementation of a relatively detailed list of reforms. As a result, Ukraine can only be obliged to take steps that do not require the enactment of new laws,14 such as presidential actions to promote reforms, unless the Verkhovna Rada ratifies the agreement.

The wording in most of the other agreements suggests that the signatories intended to frame them as political agreements, although there is some ambiguity in that regard due to the lack of universal international criteria for differentiating between binding and nonbinding agreements. Terms such as “participants,” “come into effect” (instead of “into force” like in the U.S. agreement), “determined” or “decided” (instead of “agreed”), and the verbs “will,” “would,” and “intend” (instead of “shall”) are generally viewed as nonbinding in international law.15

Also, most agreements (including the U.S. one) use the term security “commitments” instead of the more clearly treaty-like term “obligations.” The German agreement uses a less formal term, “arrangements,” which more clearly indicates that the commitments are political in nature. In Ukrainian, “commitments” and “obligations” are both translated as zobov'yazannya. As a result, the Ukrainian versions of the agreements may lead readers to have an inflated impression of the partners’ commitments. Such meaningful differences recall the strategic use of translation in the Budapest Memorandum, where the term “assurances” (in English and French) differed from “guarantees” (in the equally authentic Ukrainian and Russian texts) in order to create ambiguity around the memorandum’s legal form and to sell the deal to the Ukrainian parliament and public. Likely to avoid similar misunderstandings, most agreements (except the Canadian, French, and German ones) specify that their English versions prevail. Nevertheless, the Ukrainian public and media may still interpret the agreements as constituting “obligations.”16

The Russian leadership understands the signatories’ intention to give only flexible political commitments to Ukraine. Russia participated in the negotiation of the Iran nuclear deal and knows too well when Western countries want to water down their commitments. Therefore, codifying the agreements in legally binding form would send a clearer message to Russia that Western partners will support Ukraine over the long haul. As this and other authors have written, the commitments would be especially strengthened by the involvement of partner countries’ legislatures in the approval of the agreements, considering the fact that long-term support for Ukraine hinges on parliamentary appropriations.

On a separate note, most agreements allow the signatories to terminate them earlier than their default ten-year term without cause upon a six-month written notice (three-month notice in the case of the Netherlands). Only the Swedish and EU17 agreements do not specify a termination option before the lapse of the ten-year period. The six-month termination notice is not the best design choice considering how easily partners could backpedal on the security framework.18 That said, partners might still incur reputational and political costs by invoking the termination provision.

The Way Forward

The bilateral security agreements Ukraine has signed with its partners contain important commitments aimed at ensuring the country’s self-defense and deterrence capacity over the long haul. Unquestionably, the agreements contain more meaningful and robust commitments than the ones in the Budapest Memorandum. Yet the texts also contain ambiguities and shortcomings that could undercut their primary purpose of deterring Russia. Considering that this security framework is envisioned to last for an undefined time beyond the end of the current war—including as an interim substitute for NATO membership—it is especially important that it rests on solid foundations.

About the Author

Mykhailo Soldatenko is an attorney in Ukraine and New York and teaches law at the National University of Kyiv–Mohyla Academy. He is an incoming doctoral student at Harvard Law School. He was previously a senior associate at a leading Ukrainian law firm, practicing international dispute resolution.


  • 1The Japan-Ukraine agreement largely focuses on nonlethal support. That said, it contains elements that mirror the other agreements, such as the future attack response mechanism.

  • 2Although the EU-Ukraine agreement does not address many elements contained in the individual bilateral agreements (for example, standards of support and capability coalitions), it still touches on EU programs of military assistance to Ukraine and the consultation mechanism in case of future aggression. Moreover, considering that EU sanctions are adopted at the union level, the EU’s sanctions-related commitments are an important development.

  • 3This wording appears without “fully” in the Portugal-Ukraine agreement. The Latvia-Ukraine agreement contains softer wording on the relevant Ukrainian capability.

  • 4An outlier among the agreements is the Dutch text. The Netherlands commits to working with Ukraine “to assist the security and defence forces of Ukraine and to restore Ukraine’s territorial integrity” (emphasis added), which could be literally construed as a commitment to assist Ukraine directly, albeit with no means specified in the text. But considering the general logic of the security agreements, the signatories almost certainly did not mean to imply that the relevant joint efforts would include collective self-defense measures.

  • 5The Polish agreement states that this standard would apply “in the event of external aggression against Poland during the period preceding Ukraine’s accession to NATO.”

  • 6QME is defined as “the ability to counter and defeat any credible conventional military threat from any individual state or possible coalition of states or from non-state actors, while sustaining minimal damages and casualties, through the use of superior military means, possessed in sufficient quantity, including weapons, command, control, communication, intelligence, surveillance, and reconnaissance capabilities that in their technical characteristics are superior in capability to those of such other individual or possible coalition of states or non-state actors.”

  • 7In 1993, Ukrainian diplomats unsuccessfully requested a similar mechanism for collective responses from the signatories of the Budapest Memorandum and even prepared a draft treaty with relevant provisions in Article 6. The signatories rejected the proposal, and the memorandum only contained a commitment to consult about the text’s shallow commitments without defined aims, timing, or responses. The new security agreements provide a more robust mechanism than the one Ukraine requested in 1993, specifying its aims, timing, and responses.

  • 8The U.S.-Ukraine agreement mentions the determination of “appropriate next steps and additional defense needs.”

  • 9Notably, the UK agreement uses the term “undertakes,” which, according to UK internal guidelines, should not be used in a nonbinding agreement as it may convey legal bindingness. Still, later in the same sentence it adds the word “would,” which somewhat softens the wording. Other partners use wording that usually expresses commitments of a lesser intensity, such as “affirm” (Italy, Germany, Portugal), “intend” (the Netherlands and Japan), and “will/would provide” (many other partners). 

  • 10The United States “has long taken the position that the inherent right of self-defense potentially applies against any illegal use of force.”

  • 11The Lithuania-Ukraine agreement uses the term “parties” like the U.S.-Ukraine agreement.

  • 12This includes the political commitment (not merely intention) to “support Ukraine in developing a modern, NATO-interoperable force that can credibly deter and, if necessary, defend against future aggression.”

  • 13Article 9(2) of the Law of Ukraine “On International Treaties” requires that, among other things, any international agreement prescribing the enactment or amendment of Ukrainian law and “political agreements . . . on friendship, mutual assistance, and cooperation”—be ratified by the Verkhovna Rada. Considering that the U.S.-Ukraine agreement envisages mutual defense assistance and cooperation, it arguably requires ratification by the Rada to bind Ukraine as a matter of Ukrainian law.

  • 14According to Article VII of the U.S.-Ukraine agreement, it shall be performed “in accordance with the respective domestic laws of the Parties.”

  • 15In this context, the auxiliary verb “will” is ambiguous because it can be used for both binding and political agreements depending on the approach of a particular state. For example, according to UK and German guidelines, “will” should be used with political commitments. However, the conspicuous exception in this respect is the use in the UK agreement of the verb “undertake” in the consultation and support mechanism, which is used to express legally binding obligations per UK guidelines. In addition, some agreements sporadically use terms like “preamble,” “termination,” and “amendments,” which are usually encountered in legally binding agreements.

  • 16For the Ukrainian leadership, it was important to avoid the term “assurances,” considering the Ukrainian public’s antipathy toward anything resembling the Budapest Memorandum. Instead, Kyiv insisted on the more impressive term “security guarantees.” Ukrainian representatives have been using this term for the G7 declaration and ensuing security agreements in large measure for Ukrainian domestic messaging purposes. There are signs that the strategy has worked: recent polls show that the majority of Ukrainians believe that the agreements will “strengthen Ukraine in the fight against Russian aggression.”

  • 17The EU-Ukraine agreement will remain in effect “as Ukraine pursues its European path” and will be reviewed no later than ten years after its signing.

  • 18By contrast, the U.S.-Bahrain security agreement signed in 2023 allows termination only after the agreement has been in force for five years. A similar provision in the agreements with Ukraine would make them more credible.

Carnegie does not take institutional positions on public policy issues; the views represented herein are those of the author(s) and do not necessarily reflect the views of Carnegie, its staff, or its trustees.