The Architecture of Cruelty

The Case-Zablocki Act, 1 U.S.C. § 112b, requires the executive to transmit to Congress and publish for the public the texts of international agreements and some nonbinding instruments. Third country removal arrangements satisfy this definition because they are concluded with foreign governments and reflect a commitment by another state to receive noncitizens removed from the United States.

Under the Act, the Secretary of State must make the full text of an international agreement and any related materials publicly available no later than 120 days after the agreement enters into force, unless an exception applies. Despite these disclosure requirements, at the time of writing, not every third-country removal agreement has been published, and nearly every operating plan has been withheld.

As a result – and in violation of the law – the public lacks access to critical information about these arrangements, including, in some cases, the number of people who may be transferred, the criteria receiving states use to accept or reject refugees and migrants for removal, the procedures that will govern post-transfer treatment, and the financial or logistical support provided by the United States.

The non-disclosure of removal agreements and implementation plans, and reliance on the use of MOUs and diplomatic notes, may reflect an intent on the part of the Trump administration to end-run the Case Act.

Close review of all publicly available agreement texts spotlights when and where receiving countries have negotiated their own terms and conditions. The word "discretion" is specifically used to indicate where the receiving country has final authority.

The agreements give receiving states varying degrees of discretion over both who may be transferred and what protections, if any, people who have been transferred receive after arrival. This discretion manifests in two ways. First, many agreements allow receiving states to limit, screen, accept, or reject migrants proposed for removal based on numerical caps, nationality, age, medical history, criminal history, professional skills, or other criteria. Second, once a transfer is completed, the agreements leave the receiving state responsible for determining the transferee's legal status, access to protection, length of stay in the receiving country, and treatment under domestic law.

This structure is central to the agreements' design. The United States retains control over initiating removals, proposing specific refugees and migrants for removal, and completing the transfer. But after removal, the agreements shift responsibility for ongoing protection or transfer, status, and services onto receiving countries. The absence of any further oversight by or reporting requirements to the US indicates that the US is drawing a line once the transfer occurs.

In short, the United States limits its own obligations by relying on receiving countries that may not, or demonstrably do not, have the capacity, legal infrastructure, or political will to guarantee meaningful protection.

The agreements devote noticeably no recognition of the rights of the people transferred to the third country.

While the agreements lack explicit U.S. commitments, there are a few consistent sentences across the publicly available documents that frame the receiving state's obligations under international human rights law as commitments.

As a result, recent cases of removals in accordance with this policy demonstrate the significance of "assurances" to the whole third country removals system. If a country "assures" the United States it won't persecute people, then the U.S. will send refugees and migrants to these countries without a hearing.

Most agreements assert that the parties will act in accordance with the countries' respective international obligations.

In practice, many of these states lack the legal mechanisms to deliver on those promises. Without implementing laws and procedures, a country cannot provide what the agreements explicitly promise. The protection systems of the receiving states, are thus foundational to the agreements themselves, and the legal systems of every receiving country bear on whether migrants' fundamental human rights will survive the transfer.

Assessing those domestic regimes makes it clear that some receiving countries possess neither the legal framework nor the institutional capacity to honor their commitments, either under the agreements or under international law.

Many receiving countries have asylum or refugee protection laws on the books but no procedures to implement them. For example, Cabo Verde does not have a functioning asylum system, despite having a national asylum law. Although Law 99/V/99 defines a legal regime for asylum and refugee status, no institutions or procedures implement it; there is no refugee-status-determination mechanism, no system to register or process claims, and no designated authority responsible for asylum seekers.

Equatorial Guinea presents a similar gap. It recognizes a right to asylum in its 1991 Fundamental Law and has acceded to key instruments, including the 1951 Refugee Convention and its 1967 Protocol, the 1969 OAU Convention Governing the Specific Aspects of Refugee Problems in Africa, and conventions on statelessness, yet it has no system to receive, process, or protect asylum seekers. Despite these significant legal shortcomings and documented cases of human rights abuses, in 2025, the United States agreed to directly pay Equatorial Guinea $7.5 million in exchange for Equatorial Guinea's acceptance of third country nationals removed from the U.S.

Despite decades of U.S. immigration law guaranteeing protections for migrants, and despite nearly every agreement containing assurances to uphold the principle of non-refoulement and/or the obligations set forth in key conventions, many individuals transferred under these agreements have been subjected to the forcible return to their home countries, or countries where they face persecution. The problem is not merely that the agreements fail to prevent this risk but the receiving countries are actively violating these legal norms.

The United States began transferring West African refugees and migrants to Ghana in September 2025. Within roughly two weeks of their arrival in the country and transfer to a military camp, Ghana removed all but one of them. Among those sent onward were two Togolese women whom U.S. immigration judges had granted withholding of removal based on credible risks of female genital mutilation.

The form and content of third-country removal agreements has evolved over time. These indicators demonstrate a shift away from commitments to protections and binding language, and towards nonbinding language and vague assurances.

The clearest structural change is the disappearance of formal signed agreements after 2025. Cabo Verde was also the last of the asylum cooperation agreements; every arrangement finalized since has been a removal agreement.

The State Department did not simply stop using formal signed agreements. It also stopped using the terminology that its own Guidance associates with binding international agreements and adopted the terminology recommended for nonbinding documents. Changes in form, title, self-reference, and publication demonstrate that the United States has shifted from a mixed practice of binding and nonbinding arrangements to exclusive reliance on nonbinding instruments.

This shift towards nonbinding language has practical significance because it weakens receiving states' commitments to migrants and refugees transferred under such agreements. The March 2025 Honduras ACA uses the strongest language: "Honduras shall not return or expel a Protection Applicant" before a final administrative decision on the individual's protection request. By October, Equatorial Guinea "assures" the United States that it "intends to treat" transferred third-country nationals consistently with its obligations. The sequence moves from a mandatory prohibition, to a stated future commitment, to an assurance of intent, and finally to an assurance framed in conditional terms.

The vocabulary used to describe the arrangements has also changed. Early agreement titles framed the policy as deportation, removal, or the examination of protection claims. Later agreements increasingly use the term "transfer." Costa Rica's 2025 agreement concerned "assistance for the deportation of irregular migrants," while its 2026 agreement concerned the "transfer of third-country nationals."

In the first few months of the second Trump Administration, it quickly became clear that the "America First" foreign policy strategy champions what other countries can do for the U.S.. Early implementation of visa restrictions, travel bans and the gutting of funding for critical humanitarian aid around the world immediately placed the United States in an advantageous negotiating position with governments that were suddenly facing the disappearance of billions of dollars in promised aid.

Against this backdrop, the State Department quickly sent envoys and required ambassadors across the globe to negotiate new third country removal agreements. The department negotiated using a combination of templates and unique texts that have evolved over time, requesting the acceptance of refugees and migrants in return for financial and humanitarian aid.

Several foreign governments have expressly stated feeling pressured to sign a removal agreement to get back financial assistance. Others have alluded to promises of lifting visa restrictions in the United States or suggestions by US representatives that the acceptance of third-country nationals was needed to strengthen diplomatic relations.

Twenty-four agreements disclaim any obligation to provide funding. The absence of funding terms in an agreement does not indicate an absence of U.S. funding. The Equatorial Guinea agreement, for example, makes no reference to funds, yet the United States paid the government of Equatorial Guinea $7.5 million directly in connection with the removal agreement.

Three of the countries who explicitly agree to accept persons "who raise national security or public safety concerns" received the largest grants, totalling over $100million. These grant amounts reframe the meaning of clauses in removal agreements that disclaim any funding obligations. Therefore, language of this kind should not be read as evidence that the arrangement was unfunded or funding commitments were not integral to negotiations.