Lost for words: The dangers of AI translation & interpreting in U.S. immigration control
Washington D.C., Chicago, Los Angeles, Memphis… As the National Guard pours down on major US cities and ICE raids flood neighbourhoods, a new storm is brewing behind the scenes of the United States Department of Homeland Security (DHS): the federal use of Artificial Intelligence (AI) for translation and interpreting purposes in the context of immigration control and the dangers it poses to both citizens’ and non-citizens’ rights and safety while on US soil.
In April 2025, the Office for Management and Budget (OMB) under the Trump administration released a new memo addressing the use of AI within DHS operations, a document that some lawyers have criticised for omitting key ‘rights-impacting’ and ‘safety-impacting’ minimum risks standards, now reduced to a single, less extensive ‘high-impact’ category. These standards appear blatantly more comprehensive in the previous March 2024 memo released under the Biden administration, for example in the case of the use of AI in the screening and identification of individuals within the context of services such as public housing and education.
According to the American Immigration Council, 27 out of 105 AI use cases mentioned in the 2024 DHS inventory – over 25% – are labelled as ‘rights-impacting’. Out of those 27, at least four cases can be directly linked to the rights-impacting use of AI in translation and interpreting in three notable areas of the DHS, involving the Transport Security Administration (TSA), the Customs and Border Protection mobile application (CBP Home App), and the US Citizenship and Immigration Services (USCIS) Translation Services:
TSA: Currently at a stage of ‘pre-deployment’ and therefore not yet in action, the TSA is investigating the use of body-worn cameras that incorporate AI in their underlying software, Axon Capture Application, in order to provide transcription, translation and categorisation of audio and video data.
CBP Home App: Although awaiting further compliance checks, CBP Home is now home to two cases of AI use in translation that actively impact the rights of those on US soil. This includes the use of Babel Street, a provider of open-source threat intelligence tools used to screen and translate online information on travellers to the USA in order to identify those who may be subject to further inspection, as well as the use of AI in the CBP Translate app, used by Customs and Border Protection Officers in order to ‘expedite questioning and reduce reliance on translators to save time and costs’.
USCIS Translation Service: Initiated on 1st January 2025, this includes the use of AI-powered tools for document translation and real-time interpreting in areas such as immigration law, visas, IDs, Birth Certificates, etc. in order to ‘reduce or eliminate RSC (Resettlement Support Centre) casework dedicated to document translation.’
Building on this, there are also at least two records of the use of AI for translation and interpreting by ICE’s Enforcement and Removals Operations (ERO), however the DHS inventory doesn’t currently consider them as ‘rights-impacting’ as it is argued that they use AI to analyse ‘legally acquired email data’ and to ‘assist with real-time translation operations in communication with non-citizens in non-critical conversations’. Here the DHS assures that in the case that an individual’s rights are threatened or that the translations are unclear, professional assistance will be sought, and that in all aforementioned departments any AI use case is reviewed and approved by professional staff. However, considering contemporary criticism and accusations surrounding ICE’s alleged abuses of judicial power, who’s not to say that similar steps are not being taken in this field?
Even if it could confidently be said that all AI cases are cross-checked by professional input, the use of AI translation and interpreting in immigration control remains a major area of concern, particularly regarding transparency.
When it comes to informing the general public of the active use of AI in individual immigration and citizenship cases, the DHS still lacks concrete action, meaning that the vast majority of those affected aren’t even aware that AI is playing an active role in their case. On the same topic, there is also little to no accessible information regarding client rights in the case that AI is involved: can one opt out of having their case reviewed by AI? Is there an appeal process for those unsatisfied with an immigration court case decision influenced by AI? Is it even disclosed to the applicant, at any stage of the process, that the USCIS may use AI to translate official and personal documents, translations that may be vital to the very outcome of their case?
As long as these key questions remain unanswered, the use of AI in such instances – which can, and does, result in life-changing consequences for those on the receiving end – is highly objectionable.
Translation, as a practice and profession, is not only highly technical but also heavily nuanced, whether that may be by socio-political context, culture, idiomatic language and expressions, or even emotion. These nuances are crucial in the sphere of immigration law but run the risk of being blatantly disregarded, due on the one hand to the preconceived bias and eurocentrism that inevitably comes with the use of current AI technology, and on the other to the introduction of policies – such as those set by the DHS – that intentionally seek to target the eradication of professional translation as we know it.
Clients requiring translation or interpreting services in immigration cases are often in an extremely vulnerable position, their linguistic limitations an easy target that can quickly be weaponised by the opposition. In these cases, professional translators are more than just a service: they are a source of compassion, empathy, and hope. They are peers, bridging an invisible gap. They become a reassuring presence to a client who can finally feel understood, not only linguistically but also for who they are, what they’ve experienced and what they need moving forward.
Professional translators are many things, but perhaps most importantly – or at least for the time being – they are human.
Image courtesy of the Miami Herald via Getty Images, ©2020. Some rights reserved.
The views and opinions expressed in this article are those of the author and do not necessarily reflect those of the wider St. Andrews Foreign Affairs Review team.
