On 27 October, Open Rights Group, a UK digital rights campaign organisation, published Automated Discrimination: Unpacking the Digital Hostile Environment, a report alleging extensive AI use in the Home Office’s immigration work, the Guardian reported. The tools described range from asylum casework to border surveillance. Around 7,000 AI summaries of asylum interviews are generated per month on average, the Independent reported.
Key points
- Open Rights Group says AI is used to process asylum information, assess and prioritise immigration cases, inspect phones and monitor borders.
- Internal Home Office spot-checks found that an AI tool for summarising asylum interviews failed basic accuracy tests, the Independent reported.
- The Home Office says trained caseworkers decide asylum claims and cannot use the AI tools in isolation.
- The group wants tighter regulation, greater disclosure and a public register of government AI tools.
AI across Home Office immigration work
Open Rights Group assembled its account from limited Home Office publications, freedom of information requests and questions asked by MPs. Its report says officials have routinely relied on national security grounds to resist requests for information and redact material they release. Labour’s Bell Ribeiro-Addy, the Scottish National party’s Pete Wishart and the Green party’s Siân Berry support the report.
According to the group, AI is used to check the contents of phones belonging to new arrivals and to flag marriages that may be sham arrangements. Its account also describes border monitoring using live facial recognition, AI-powered drones and satellites. Tools are being developed to predict the ages of people arriving by small boat from their facial features. Open Rights Group describes AI use in determining the ages of asylum seekers.
Within asylum processing, the report describes AI-generated summaries of interviews and of country policy information used as evidence. It also identifies the Identify and Prioritise Immigration Cases tool, or Ipic, as a means of assessing and ordering immigration cases. These are different points in a case’s progress: what information a caseworker sees, how it is condensed and which cases receive attention. The Home Office says the tools assist staff rather than decide claims.
The group places these uses within the “hostile environment”, the approach to immigration first envisaged by Theresa May as home secretary in 2012. Open Rights Group argues that AI can reproduce discrimination and objects to treating individual immigration cases as technical problems. Sara Alsherif, a programme manager at the organisation, called for transparency, accountability and restrictions on further use.
Around 7,000 asylum summaries each month
The scale of the interview-summary system makes the quality of its output consequential for routine casework. Around 7,000 summaries are generated per month on average, although individual decision-makers can choose whether to use them. Home Office spot-checks found that the tool failed basic accuracy tests. A faulty account of an interview could therefore be available to a caseworker even where the decision remains theirs.
The Home Office says trained caseworkers make asylum decisions. Its spokesperson said AI assists them in finding their way through interview records and policy information, cannot be used on its own and does not decide claims. The department says it monitors the tools through its existing quality-assurance processes and has committed to publishing updated evaluations of their rollout.
Ribeiro-Addy wants the Home Office to publish its equality and data protection assessments, explain staff training and permit independent checks of the tools’ performance. She said asylum seekers are not told when AI has been used to summarise their cases. Her objection goes beyond whether a caseworker signs off the result: a person contesting an account of their interview needs to know how that account was produced.
Lawyers at Cloister Chambers and Doughty Street Chambers have argued that the Home Office’s AI use is “likely to be unlawful” because, in their view, it falls short of legal obligations and standards in the government’s AI playbook, the Independent reported earlier in 2026. The issues they raised include public transparency and consideration of alternatives before adopting the tools. That is a legal argument about the department’s practices, rather than a court ruling on them.
Open Rights Group calls for a register
Open Rights Group wants the government to regulate AI and prohibit uses it considers undemocratic or hazardous to human rights, including the use of ChatGPT tools in asylum applications. It also calls for stronger freedom of information laws, a narrower national security exemption and a public register of government AI tools. Those requests address how people outside the Home Office learn which systems are operating, as well as the conditions under which they may be used.
Wishart said an asylum seeker who believes an AI system has made an error should be able to learn that it was used and contest its output. Berry called for laws grounded in human rights and fairness, with decisions open to scrutiny and challenge. Ribeiro-Addy asked the department to allow independent validation of how well its tools perform.
