UK Watchdog Weighs Probe Into Home Office AI Deployment
AI-generated image · US National Wire
A government inspector is considering an investigation into the use of OpenAI-powered tools to process asylum claims amid reports of factual errors and a lack of transparency.
The Independent Chief Inspector of Borders and Immigration (ICIBI) is considering a probe into how the UK Home Office uses generative AI to process asylum applications, as Wired first reported.
The Home Office deployed two tools—Asylum Case Summarisation (ACS) and Asylum Policy Search (APS)—built on Microsoft Foundry and OpenAI’s 2023 GPT-4 model. Launched in late 2024 to address a backlog of approximately 100,000 cases, the tools are designed to summarize interview transcripts and country policy information. Government estimates suggest the tools can save roughly one hour per case.
However, the nonprofit Open Rights Group (ORG) conducted a six-month investigation involving public records requests and parliamentary questions, finding that the ACS error rate is significantly higher than the 9 percent reported in pilot evaluations. ORG also noted a total lack of quality control for the APS tool. In July, a UK judge accused the Home Office of using an "AI hallucinated" document to deny an asylum claim for a Moroccan woman and her child.
Critics, including Labour MP Bell Ribeiro-Addy and the Public Law Project, have raised concerns regarding cultural bias in GPT models and the potential for discriminatory decisions. Furthermore, the tools are not listed on the public repository required by the UK’s "AI Playbook," and asylum seekers are not notified when AI is used in their cases.
A Home Office spokesperson told Wired that AI does not replace human decision-makers and is used only to assist caseworkers. An OpenAI spokesperson noted the company is working to improve accuracy and will retire GPT-4 in January.