A UK government watchdog is weighing whether to investigate the Home Office's deployment of generative AI tools to handle asylum applications, according to reporting by WIRED. The Independent Chief Inspector of Borders and Immigration, which oversees the Home Office's border and immigration management, has requested additional details from digital rights nonprofit Open Rights Group about the tools as it plans next year's inspection topics. The potential probe comes after lawmakers and researchers raised public concerns about the tools' reliability and the government's lack of transparency surrounding their deployment.
The Home Office uses two AI systems—Asylum Case Summarisation and Asylum Policy Search—built on OpenAI's 2023 GPT-4 large language model and Microsoft Foundry, to condense interview transcripts with asylum seekers and information about their countries of origin, according to documents obtained by Open Rights Group. The immigration authority began testing these tools in late 2024 as the Labour government worked to eliminate a backlog of roughly 100,000 asylum cases while managing tens of thousands of new applications the UK receives annually. While the department cut the number of people awaiting an initial decision by half in the year through June, the volume of appeals against decisions has surged. The pilot showed the tools could reduce processing time by approximately one hour per case when used in combination, based on government calculations.
Open Rights Group's investigation, which spanned six months and included 11 parliamentary questions, two ministerial letters, and 10 public records requests, uncovered that the share of summaries generated by the Asylum Case Summarisation tool judged unsuitable for use—frequently due to factual mistakes or distortions—is substantially higher than the published 9 percent error rate from the pilot evaluation, the nonprofit found. The Asylum Policy Search tool has no quality control measures for its generated summaries, the group also discovered. The watchdog noted in previous findings that one tool's "effectiveness and reliability were varied," while a Home Office spokesperson stated that "asylum decisions are made by trained caseworkers, with AI tools used only to assist in navigating interview records and policy information."
Researchers and lawmakers worry that well-documented cultural bias in GPT models could create language obstacles and erase cultural subtleties in the summaries, potentially producing discriminatory outcomes, according to the report. Asylum seekers aren't informed when these AI tools are applied to their cases, and migrant organizations and legal practitioners remain largely unaware of the tools' actual use, according to Sara Alsherif, who leads Open Rights Group's migrant digital justice program. Independent legal analysis commissioned by the nonprofit concluded that preventing asylum seekers from challenging an incorrect AI summary of their personal information may violate UK data rights laws. The Public Law Project identified the two Home Office tools as part of 11 high-risk AI systems being used across government but missing from the public repository that UK's AI Playbook requires departments to maintain. The broader concern extends beyond accuracy to accountability—government may be deploying additional AI tools that even researchers can't identify, let alone the individuals whose lives they affect. For decision-makers navigating the governance of emerging technology, the tension between operational efficiency and procedural fairness will only sharpen as automation scales faster than oversight frameworks can adapt.