A UK watchdog may investigate the Home Office's use of ChatGPT-based tools in asylum cases, over doubts about accuracy and a lack of openness.
A UK government watchdog is considering an investigation into how the Home Office uses ChatGPT technology to help process asylum claims. The Independent Chief Inspector of Borders and Immigration has asked the digital rights group Open Rights Group for more information as it plans next year's inspections. The watchdog says it cannot confirm its topics until it publishes its priorities.
The Home Office uses two tools. One summarizes transcripts of interviews with asylum applicants. The other works as a search assistant for information about an applicant's home country. Both are built on OpenAI's 2023 GPT-4 model and Microsoft Foundry. Piloting began in late 2024, as the government tried to clear a backlog of around 100,000 cases. Together, the tools could save about an hour per case, according to government estimates.
The Home Office's own pilot put the error rate for the interview summaries at 9 percent. Open Rights Group says the share of summaries judged not fit for use is significantly higher, often because of factual errors or misrepresentations. It also found no quality control for the country search tool.
In July, a UK judge accused the Home Office of citing an "AI hallucinated" document when refusing a Moroccan woman and her child asylum. Researchers also worry that known cultural bias in GPT models could strip nuance from summaries and lead to unfair decisions.
UK government guidance says departments must be open about algorithmic tools used in decisions, including by listing them in a public repository. These two tools are not on that list. Asylum seekers are not told when AI is used in their case.
A spokesperson says trained caseworkers make asylum decisions. The AI tools only help them navigate interview records and policy information, and they cannot be used alone or decide claims. The department says it monitors the tools and has committed to publishing updated evaluations. Its privacy notice says parts of the process may involve automation.
The watchdog has not said whether it will investigate. Open Rights Group and the Public Law Project have sent their findings to the UK data regulator, the Information Commission. It is already investigating the Home Office's refusal to say how staff use and were trained on the tools. OpenAI says it will retire GPT-4 in January.
When AI summaries feed into decisions about people, those affected should be told, and the error rates should be public and checked.
Code in an iOS 27.3 beta suggests Apple's HomeView smart display could run AI chatbots such as ChatGPT next to Siri. Nothing is confirmed.
Read more: Apple's smart home screen may soon let you talk to ChatGPTA How-To Geek writer has Claude read his Todoist list each weekday morning and brief him, with read-only access and no changes unless he says yes.
Read more: He set up Claude to read his to-do list and brief him every morningBloomberg reports Apple's CEO is pushing the head of its software team to take bolder steps on AI, which could shape features on your iPhone and Mac.
Read more: Apple's CEO pushes its software chief toward bolder AI plans
