UK watchdog weighs probe into Home Office's OpenAI-based asylum tools
A government inspector is considering investigating the Home Office's ChatGPT-based tools after a digital rights group found error rates in asylum case summaries higher than officially reported.
What to know
- ICIBI, the UK's borders and immigration watchdog, is weighing an inspection of the Home Office's OpenAI-based asylum tools after requesting more information from Open Rights Group.
- ORG's six-month investigation found the error rate in AI-generated case summaries is significantly higher than the Home Office's published 9 percent pilot figure, and that one of the tools has no quality control.
- The Home Office has withheld details on the tools despite 11 parliamentary questions, two ministerial letters and 10 public records requests, and maintains the AI only assists caseworkers who make the final decisions.
- The tools, used since late 2024 to help clear a roughly 100,000-case asylum backlog, carry high stakes since decisions affect whether applicants are returned to countries where they fear persecution.
“effectiveness and reliability were varied”
ICIBI, UK government watchdog · WIRED ↗ · Oct 26
Sara Alsherif Open Rights Group campaignerOpen Rights Group Digital rights nonprofitIndependent Chief Inspector of Borders and Immigration (ICIBI) Government watchdogUK Home Office Government department using the tools
How it unfolded 2 developments, newest first · click a bar or a number to jump articlesposts
-
2
Open Rights Group publicizes probe and its findings
ORG posts on Mastodon highlighting the possible Home Office probe and restating that its investigation found errors and inaccuracies in the ChatGPT-based tools used to summarize interviews and policy documents.
“The UK Home Office may face a probe by a government watchdog 🚨 This comes following the findings in ORG's six-month investigation into the use of AI in asylum decision-making. We found errors and inaccuracies in the ChatGPT tools being used to summarise interviews and policy documents.”
— Open Rights Group -
O
The UK Home Office may face a probe by a government watchdog 🚨 This comes following the findings in ORG's six-month investigation into the use of AI in asylum decision-making. We found errors and inaccuracies in the ChatGPT tools being used to summarise interviews and policy documents. Read more ⤵️ https://www. wired.com/story/uk-home-office…
-
- 2 days quiet
-
1
ICIBI weighs inspection of Home Office's AI tools, Wired reports
WIRED reports the Independent Chief Inspector of Borders and Immigration has asked ORG for more information on the tools as it plans next year's inspection topics; the Home Office defends the tools as assistive only, not decision-making.
“Asylum decisions are made by trained caseworkers, with AI tools used only to assist in navigating interview records and policy information. They cannot be used in isolation and do not make decisions on claims…”
— Home Office spokesperson -
first by Mastodon, 17d ago · also Wired
1 more headline
- UK Home Office May Face Investigation Over Use of OpenAI Tools Wired · 17d ago
-
C
https://www. wired.com/story/uk-home-office -may-face-investigation-over-use-of-openai-tools/
-
-
background
Open Rights Group completes six-month probe into the tools — ORG's investigation, involving 11 parliamentary questions, two ministerial letters and 10 public records requests, finds the share of ACS summaries deemed not fit for use is significantly higher than the published 9 percent pilot error rate, and that APS summaries have no quality control.
-
background
Home Office begins piloting OpenAI-based asylum tools — The Home Office starts piloting the Asylum Case Summarisation (ACS) and Asylum Policy Search (APS) tools, built on OpenAI's 2023 GPT-4 model and Microsoft Foundry, to help clear a backlog of around 100,000 asylum cases.