UK Home Office faces probe over OpenAI tool use in asylum claims
uk home office may face investigation over use of openai's gpt-4 in asylum processing amid concerns over accuracy and secrecy. this affects asylum seekers and raises questions about ai reliability in high-stakes decisions.
the uk home office has been using openai's gpt-4 model, specifically the asylum case summarisation and asylum policy search tools, to process asylum applications since late 2024. these tools aim to help caseworkers navigate interview records and policy information, but have raised concerns over accuracy and secrecy. a government watchdog, the independent chief inspector of borders and immigration, is considering an investigation into the home office's use of these tools. researchers found that the percentage of summaries generated by the asylum case summarisation tool deemed not fit for use is significantly higher than the published 9 percent error rate. the probe follows lawmakers and researchers questioning the tools' accuracy and the government's secrecy around their use.