UK secrecy over AI in immigration and asylum decisions under fire

The government is using AI in “near total secrecy” to police immigration and border crossings and is routinely invoking “national security” to refuse to answer questions on its use of automated technology, it is claimed.

MPs and campaigners warned today that secrecy around the Office’s use of AI in asylum and immigration meant there was little accountability over the UK’s use of the technology or its effectiveness.

Migrants and people crossing into the UK were at risk of discrimination, biased decision making and of being wrongly identified by facial recognition systems, according to research by the Open Rights Group.

The campaign group warns in a 100 page report that the government is using “technical fixes” – what it calls “tech no-solutionism” - to implement policy decisions around “the complex human phenomenon of migration”.

The Office has embedded AI into systems used to monitor borders through live facial recognition and is using AI-powered drones and surveillance towers to monitor small boat crossings.

It is also using AI to estimate the ages of people seeking asylum, to summarise case interviews of asylum seekers, to analyse the contents of mobile phones, to automatically identify sham marriages, and to prioritise immigration cases.

UK locked-in to overseas technology

Most of the technology used is supplied by US, Israeli and other foreign defence and security companies, leaving the UK potentially locked-in to overseas technology, rather than developing its own AI solutions.

The report warns AI can make mistakes, produce inaccurate information, leave out important details, or present wrong information in a way that makes it appear reliable with “ruinous” consequences for asylum seekers.

Computer Weekly previously reported that the Office’s Asylum Case Summarisation (ACS) tool, which is used to summarise interviews with asylum seekers, had been deployed despite a significant proportion of summaries failing quality and accuracy checks .

Another algorithmic tool, Identify and Prioritise Immigration Cases (IPIC), automatically recommends migrants for actions, including detention and bail conditions.

According to research by the campaign group Privacy International, the algorithm is designed to push case workers into accepting its recommendations, by requiring officials to provide an explanation if they disagree.

IPIC uses nationality to inform its decisions, leading to risks that people with certain nationalities could be more likely to receive negative decisions, such as being recommended for detention or electronic tagging, the report claims.

Another system, the Electronic Monitoring Review Tool (EMRT), automatically generates “harm scores” to determine how long people should wear ankle tags before being moved to other forms of monitoring.

The Office uses the Atlas platform to manage the casework for immigration and asylum applications, using automated tools to select and deliver data to caseworkers for decision making.

Leaked documents reported by the Guardian revealed that flaws with the database led to more than 76,000 people being listed with incorrect names, photographs or immigration status.

Published contracts reveal that Atlas is due to be enhanced with AI, as part of an £87 million deal with Mastek, an IT service provider with headquarters in India, though no details have been disclosed.

Under another Office contract , Mastek is providing AI to enhance the government’s Biometric Services Gateway and National DNA database for immigration and asylum purposes.

The Office is trialing AI to estimate the ages of people seeking asylum, after awarding a contract for facial age estimation (FAE) in May to UK-based Akhter Computers and Germany’s Cognitec Systems.

The trials, expected to be carried out at the Western Jet Foil processing centre in Dover, will focus on 16–18-year-olds, but even the best technology has an error margin of 2.5 years, creating risks that juveniles will be wrongly identified as adults.

Human rights groups warn that the technology has more difficulty estimating the ages of women and people of colour, and that children seeking asylum often appear prematurely aged due to trauma, violence, lack of food and sleep and exposure to salt water during dangerous sea crossings.

Mobile phone data extraction

In January, new powers in the Border Security Asylum and Immigration Act 2025, came into force allowing immigration officers to extract data from asylum seeker’s mobile phones, with trials beginning in Manston migrant processing centre in Kent .

Extracting phone data is likely to use tools from suppliers such as the Israeli company Cellebrite, which uses AI to analyse call records, text messages, images and videos stored on mobile phones and other electronic devices at high speed.

The Office is using a machine learning tool to automate the detection of sham marriages, based on six undisclosed risk factors, which according to the report, has disproportionately selected a limited range of nationalities – despite any evidence that they are more likely than other nationalities to be involved in sham marriages.

The UK has taken part in EU-funded projects aimed to develop powerful surveillance capabilities at borders. They include CarMen, which promises the real-time biometric identification of passengers in vehicles moving up to 20 km/h and PopEye which aims to identify multiple travellers crossing a border on foot or in a vehicle from up to 200 meters away.

Multiple pilots of live facial recognition are being conducted at Holyhead in Anglesey, Wales, to identify people returning to the UK in breach of deportation orders.

A trial in April scanned thousands of faces of people traveling from Dublin to Holyhead but produced no matches, raising questions the effectiveness and proportionality of the technology, according to a report by the Irish Times .

Anduril Sentry Towers

The Office’s use of AI-powered surveillance towers was revealed by journalist and researcher Samuel Storey . The towers, supplied by US defence firm Anduril, use an AI-based operating system to produce a real-time picture of the border and to identify and track individuals of interest.

The Office put Storey under pressure to remove details of the locations of the towers he had identified, claiming both that they were inaccurate and likely to lead to countermeasures by criminals.

Investigations have also revealed that the Office uses drones produced by foreign security companies, including Tekever in Portugal and Israel’s Elbit Systems to patrol the channel to identify small boats. Computer Weekly previously reported Office plans to deploy AI-powered satellites to counter small boat crossings .

Call to address gaps in regulation

The Open Rights Group calls for the government to address gaps in regulation of AI systems, including banning the use of mass live surveillance in public places and regulating high-risk applications of AI.

There should be transparency government use of AI, including a robust public register of AI tools, and a rejection of “AI techno-solutionism” it said.

Labour MP Bell Ribeiro-Addy said that the Office had been reluctant to publish documents to explain how AI is used in asylum processing.

“The Office has not explained how it will ensure there is accountability when they fail or justified why some of the most vulnerable people, with the least power, should bear the greatest risks,” she added.

SNP MP Pete Wishart criticised the Office’s failure to inform asylum applicants when AI is used in their cases, meaning they could not challenge their decisions.

“There is also an important question of accountability. If an asylum seeker believes that an AI system has made a mistake, they need to be able to know that the technology was used and have the chance to challenge its output,” he added.

Green Party MP Sian Berry said that testing AI on migrants has consequences for the rest of society including normalising biometric checks, digital identification, and sharing data on individuals.

“Artificial Intelligence systems must be subject to clear laws founded in human rights and fairness, with decisions subject to scrutiny and challenge. Human beings need to make decisions people, and not rely on opaque calculations,” she added.

She said that public bodies should not hand over democratic control of the way technology is designed and deployed to “a handful of corporations with dubious human rights records.”

Claudia Aradau, Professor of International Politics, King’s College London, said that AI systems amplified biases in the data used to train them, and created a hostile environment for immigrants.

“These systems do not just amplify existing social biases in the data but produce hostility as both an intended and unintended consequence,” she said.

AI tools do not replace humans

A Office spokesperson said that Artificial Intelligence is just one of the tools available and does not replace or automate the role of human decision-makers.

“Asylum decisions are made by trained caseworkers, with AI tools used only to assist in navigating interview records and policy information. They cannot be used in isolation and do not make decisions on claims,” the spokesperson added.

“We regularly monitor the performance of these tools alongside our established quality-assurance processes, and have already committed to publishing updated evaluations on their roll out.”

AI tool used in asylum case reviews failing accuracy tests - Asylum seekers are not informed that caseworkers use an AI tool to summarise their interviews used by decision makers .

UK lawmakers call for new law to protect human rights from AI systems – MPs and peers call for the UK to create a statutory AI oversight body with powers to test and evaluate high-risk uses of AI and to prevent the deployment of AI that poses risks to human rights.

Firms can’t count their AI agents, let alone switch them off – RSA’s CEO wants AI agents managed like employees, but other security leaders differ on how far that should go .

Bank of England warns of AI debt – Bank of England governor Andy Bailey warns that artificial intelligence benefits may take longer to be realised .

AI tool used in asylum case reviews failing accuracy tests By: Bill Goodwin

Amnesty: AI surveillance risks ‘supercharging’ US deportations By: Tiago Ventura

ICO investigates lawfulness of algorithms used in immigration enforcement By: Bill Goodwin

UK government outlines plan to surveil migrants with eVisa data By: Sebastian Klovig Skelton