Inviting a Wolf to the Table: The 2020 US-Taliban Agreement and Questions of State Responsibility

On Sunday 15 August 2021, towards the final phase of the US and other NATO States’ withdrawal from Afghanistan, the Taliban entered Kabul’s abandoned presidential palace, confirmed its control over the country, and announced the restoration of the Islamic Emirate of Afghanistan. The act was the culmination of a well-organized takeover by the Taliban of 26 out of 34 provincial capitals within ten days. In addition to the over 1000 civilians killed in the span of a month, enforced disappearances, and houses torched as the Taliban asserted its control in new areas, an estimated 244,000 people were forced to flee their homes since foreign forces began their withdrawal in May 2021. While the Taliban expressed its aim to form an “open, inclusive Islamic government”, the memory of the fundamentalist and violent Taliban rule in the period 1996-2001 raised alarm among the population of what the content of such “open, inclusive Islamic government” may look like.

Rather than being a mere consequence of US withdrawal, the Taliban’s takeover can be traced back to a 2020 Agreement between the USA and the Taliban. This agreement, in simple terms, handed the keys of the presidential palace over to the Taliban. Had this been a property transaction with one party handing over the keys of a house they did not own to a third party, the transaction would likely be considered void. Yet, when it came to a transaction involving a sovereign State with a population of 30 million it appears that few eyebrows were raised.

The 2020 Agreement raises interesting questions of international law. This post highlights one: the question of State responsibility. In other words, does international law prescribe any responsibility for the US when it, through the agreement, not only facilitated the arrival of the Taliban to power but actively invited it?  

To address this, let’s first take a look at the agreement.

The 2020 Agreement

On 29 February 2020, the USA and the Taliban, an armed non-state actor listed as a terrorist organization by the UN Security Council in Resolution 1267, signed what became known, euphemistically, as the “Peace Agreement”. The official title of the agreement is “Agreement for Bringing Peace to Afghanistan between the Islamic Emirate of Afghanistan which is not recognized by the United States as a state and is known as the Taliban and the United States of America”. Strikingly, for an agreement that seeks to bring peace to Afghanistan, Afghanistan itself as a State was not party to the agreement, despite the fact that it deals with issues at the core of Afghanistan’s sovereignty.

The little over three-page agreement, in addition to announcing a timeline for the withdrawal of all foreign forces from Afghanistan, includes: a commitment by the Taliban to prevent the use of Afghanistan’s territory by any group or individual against the security of the US and its allies; a ceasefire; the release of Taliban prisoners and prisoners of what is described as “the other side”; a US review of sanctions against members of the Taliban with a view to remove them; the initiation of negotiations between the Taliban and what are described as “Afghan sides”; and an agreement over the future political roadmap of Afghanistan, including the establishment of a new post-settlement “Afghan Islamic government”.

In particular the latter two points of the agreement are matters of an inherently sovereign nature as they concern the political system and governance of Afghanistan, and where the consent of Afghanistan would be a basic condition. The agreement’s reference to a future political roadmap is particularly striking as it prescribes the establishment of a new government in Afghanistan which, given its name in the agreement (“Afghan Islamic government”) signals that it will be something different from the existing “Islamic Republic of Afghanistan”. This suggests not only a political but possibly also a more fundamental constitutional change. Thus, the agreement not only explicitly invites the Taliban to negotiations about Afghanistan’s future but also, though less explicitly, envisages a role for the Taliban in the new government.

The question raised is: should the US bear any responsibility for entering into an agreement with the Taliban with such a content, without the inclusion of the Afghan government? State responsibility under international law can be triggered if an act or omission is attributable to a State and if it also breaches an international obligation owed by that State. The customary law on state responsibility is reflected in the International Law Commission (ILC) Articles on Responsibility of States for Internationally Wrongful Acts (ARSIWA). I will briefly touch upon possible grounds for responsibility.

State responsibility for own conduct?

In terms of international obligations relevant for the assessment of the US’s own conduct which may trigger state responsibility, two are particularly relevant: the prohibition under customary international law against non-intervention in the internal affairs of another State, and the prohibition against supporting terrorism and terrorist groups under UN Security Council Resolution 1373 and Resolution 1566.

It should not be controversial to claim that in entering into negotiations with the Taliban prescribing a new government and the role of the Taliban in that government, the US has interfered with the internal affairs, sovereignty, and self-determination of Afghanistan. Whether this would also constitute an intervention in violation of international law that would raise state responsibility is another question, and depends largely on the degree of coercion. As formulated by the ICJ in the Nicaragua case, “the principle [of non-intervention] forbids all States or groups of States to intervene directly or indirectly in the internal or external affairs of other States” and that “a prohibited intervention must accordingly be on matters in which each State is permitted, by the principle of State sovereignty, to decide freely. One of these is the choice of political, economic, social and cultural system, and the formulation of foreign policy. Intervention is wrongful when it uses methods of coercion in regard to such choices, which must remain free ones: (…) the element of coercion (…) defines, and indeed forms the very essence of prohibited intervention” (para 205). The Court did not opine on how coercion might be defined outside use of force context, and as observed by some commentators (see for example, Jamnejad and Woods) it is unclear what is and what is not prohibited under customary international law outside the use of force context. While one may claim that the unauthorized exercise of authority in relation to another State’s exercise of independent and exclusive powers on its territory may not in itself constitute coercion, one could argue that there may be a coercive element present when a state bestows upon an armed non-state actor within the territory of another state the authority to govern. Perhaps even more so, when that armed non-state actor in question is listed as a terrorist organization. One could therefore consider that by empowering the Taliban through this agreement, the US effectively compels an outcome with respect to Afghanistan’s exercise of sovereign functions, that the intervention in effect targets the State into a change of governance, and hence entails a coercive element that would lift the level of interference above the threshold for state responsibility.

Secondly, one could consider that the content of the agreement, in particular the elements of the agreement that deal with the Taliban’s role in governance, constitutes a breach of the obligation not to support terrorism under Security Council Resolutions 1373 and 1566, especially when considering that the Security Council listed the Taliban as a terrorist organization in resolution 1267.

State responsibility for acts committed by the Taliban?

The situation also begs the more interesting question of State responsibility for acts committed by the Taliban subsequent to the agreement. Documented acts by the Taliban since 29 February 2020 include the killing of civilians, government officials, police and the army; enforced disappearances; looting and burning down of houses; detentions and other retaliatory acts against individuals, in particular those deemed to have supported the Afghan government or foreign governments.

The consideration of State responsibility in connection with these acts depends on the standard of attribution of conduct by non-state actors to States and whether, in this case, the US’s conduct through the agreement has reached that threshold.

The ICJ’s decisions in the Nicaragua and Bosnia Genocide cases, and the decision of the ICTY in Tadić, continue to shape the debate on the content of the control test for attribution: the level of control a State has to exercise over conduct by a non-state actor in order to attribute that conduct to the State. As stated unconditionally by the ICJ in the Bosnia Genocide case, the relevant test for the attribution of conduct is that of effective control.

It is clear that the Taliban is neither a de jure nor a de facto organ of the US. By “effective control”, as interpreted by the ICJ, it would be required that the US should have directed or enforced the perpetration of acts contrary to human rights law by the Taliban. This would mean either the issuance of directions to the Taliban concerning specific operations, i.e., the killing of civilians, or forcefully making the Taliban carry out those specific operations. Based on the wording of the 2020 Agreement, there are no grounds to conclude that the agreement itself constitutes such issuance of directions or forceful ordering, and hence cannot be said to reach the threshold of “effective control”.

Notwithstanding the “effective control” test, the debate has continued about a more lenient standard of attribution along the line of Judge Jenning’s dissenting opinion in ICJ’s ruling in Nicaragua, which has been reflected also in the jurisprudence of regional human rights courts (see for example, the Inter-American Court of Human Rights, Rochela Massacre v. Colombia; European Court of Human Rights, Ilascu and Others v. Moldova and Russia; and African Commission on Human and Peoples’ Rights, SERAC and Another v. Nigeria).

By pointing to a responsibility gap, where certain acts of states that contribute towards the conduct of non-state actors are not captured by the strictness of the effective control test, some commentators have argued for a lower threshold for attribution. For example, Vladyslav Lanovoy proposes that complicity (aid and assistance) should be used as a test of attribution of conduct. However, also that approach may encounter some difficulties, as noted by Ilias Plakokefalos.

One could conceive of a more lenient standard of attribution in situations where the non-state actor in question is listed by the Security Council as a terrorist organization. Instead of requiring “effective control” in the form of “direction” or “enforcement”, the engagement of States with such actors, especially bestowing them with governmental-like functions, should result in some form responsibility. The underlying idea here would be that in view of the past record of the Taliban and its established goals, as recognized by the Security Council in Resolution 1267, the empowerment of the Taliban by granting them elements of government authority, would with a high likelihood be expected to be exercised in ways that would lead to gross violations of international law. Establishing State responsibility in these situations could have a preventive effect, and be more consistent with the purpose of the Security Council’s listing of some non-state actors as terrorist groups. It would also be more consistent with the parallel development towards allowing States to use force against terrorist groups in other countries. If there is a move towards a more lenient standard of attribution in the context of States’ use of force against non-state actors (see Christian Tams), it would be more consistent if the threshold for attribution was also more lenient with respect to States that empower the same non-state actors with governmental authority.