Digital-rights campaigner Open Rights Group (ORG) has published a 20-page report condemning two artificial-intelligence tools now embedded in the UK asylum workflow: the Asylum Case Summarisation (ACS) system and the Asylum Policy Search (APS) engine. ORG’s six-month investigation found ACS summaries were deemed “not fit for use” in 22 % of sampled cases, far higher than the 9 % error rate disclosed in the Home Office’s own pilot. The search tool, which draws on GPT-4o through Azure AI, has no ongoing quality-control process. ORG argues that the Home Office ignored the UK government’s AI Playbook and failed to list the systems on the Algorithmic Transparency Recording Standard hub. Parliamentary answers and freedom-of-information responses provided contradictory accounts of staff training, prompting calls for an independent audit by the Chief Inspector of Borders and Immigration. For employers sponsoring refugee talent or advising global assignees on protection routes, unreliable AI-generated summaries could delay substantive decisions—ORG calculated a 29 % increase in the time between interview and decision since April 2026. That unpredictability complicates onboarding timelines and may jeopardise travel plans conditional on fast grant of protection status. The report urges an immediate suspension of both tools, independent evaluation, and consultation with affected communities. If accepted, a moratorium could slow asylum-case throughput further in the short term but ultimately provide more legally robust outcomes, benefiting HR teams seeking certainty on right-to-work dates.
Source: Electronic Immigration Network
