A report from the Open Rights Group, covered by The Register, finds that the UK Home Office's AI tools for asylum case processing are failing accuracy checks at a high rate. The Asylum Case Summarisation system, which uses OpenAI's GPT-4o via Microsoft's Azure AI Foundry, produced 45 failed summaries out of 203 checked after its April 2026 rollout — a 22 percent failure rate, up from 9 percent in the pilot. Another 19 summaries were flagged for further review, and 30 of 56 second-stage reviews failed to meet the required standard.
The report also questions whether the tools are delivering promised efficiency gains. Home Office experts checked only 0.3 percent of summaries between April and July, based on a projected monthly caseload of 17,515. Meanwhile, the average monthly number of initial asylum decisions fell about 42 percent compared with the same period in 2025, and the mean time from interview to decision rose 29 percent. Separately, expert testing of the Asylum Policy Search tool stopped in August, despite earlier assurances to MPs that checks were continuing.
The Open Rights Group argues that human oversight may be weaker than claimed, since the summarisation tool automates work previously done by technical specialists. Applicants are not told when AI is used in their cases and cannot inspect or correct the summaries. The group is calling for both tools to be suspended and independently audited, saying current safeguards are inadequate for decisions that determine whether someone can remain in the UK.