←Story detail

Trending story

UK border inspection body considers investigating Home Office's use of generative AI

1 reports1 story updatesUpdated Oct 27 23:10

Understand this story

AI overview

The Independent Chief Inspector of Borders and Immigration (ICIBI) is considering investigating the Home Office's use of generative AI tools to process asylum applications, and has asked the digital rights organization Open Rights Group to provide more information. The Home Office's Asylum Case Summarisation and Asylum Policy Search tools are built on OpenAI's 2023 GPT-4 and Microsoft Foundry, and began piloting in late 2024. Open Rights Group's six-month investigation found that the proportion of ACS-generated summaries judged unusable was significantly higher than the 9% error rate published in the pilot evaluation, while APS-generated summaries had no quality control; these tools are also not included in the list of algorithmic tools the UK government requires to be disclosed. The investigation is currently at the consideration stage.

AI-generated from public reports

Latest development · Report

UK border watchdog may investigate Home Office's use of OpenAI tools to process asylum applications

Developments

1 confirmed developments

  1. UK border inspection body considers investigating Home Office's use of generative AI

    Source:Wired AI

Reports and discussion timeline

Browse reports and discussion by time. Their evidence levels remain distinct.

Oct 27

  1. Wired AI

    UK border watchdog may investigate Home Office's use of OpenAI tools to process asylum applications

    The Independent Chief Inspector of Borders and Immigration (ICIBI) is considering investigating the Home Office's use of generative AI tools to process asylum applications and has asked the digital rights organization Open Rights Group for more information. The Home Office's Asylum Case Summarisation and Asylum Policy Search tools are built on OpenAI's 2023 GPT-4 and Microsoft Foundry and began piloting in late 2024. ORG's six-month investigation found that the proportion of ACS-generated summaries judged unusable was significantly higher than the 9% error rate published in the pilot evaluation, while APS-generated summaries have no quality control; these tools are also not included in the list of algorithmic tools that the UK government requires to be disclosed.

Related stories

  1. Related storiesOpenAI announces frontier model training will no longer connect to the live internet
  2. Related storiesOpenAI announces ban on direct internet access for models during training
  3. Related storiesPope Leo XIV calls for disarming AI
  4. Related storiesOpenAI reports two incidents of internal coding tasks weakening misalignment monitoring