Home Office withholds AI use details from asylum claimants, minister confirms

TL;DR:

• Border security and asylum minister Alex Norris has confirmed in a written parliamentary answer that the Home Office does not provide asylum claimants with information about the AI tools used in their cases — the position is unchanged by the recent introduction of two AI-powered systems into caseworking.
• The two tools support extracting and analysing information from interview transcripts and answering caseworker policy queries; the minister said AI does not make the underlying decision, in line with a “human-in-the-loop” principle.
• Resultsense view: this lands in the same week the ICO published its draft automated-decision-making guidance and FOI guidance for AI-generated requests. The Home Office position will face scrutiny under both regimes — particularly the new Article 22C transparency safeguards that took effect on 5 February 2026.

The disclosure came in response to a written question from Labour MP Kate Osamor. Norris told Parliament: “No process and/or tooling details are currently released to asylum claimants — this has not changed with the incorporation of AI elements into caseworking.” The minister stressed that AI systems do not determine asylum outcomes, but provide analysis to inform human caseworker decisions.

What the AI tools do

The Home Office disclosed last month that two AI-powered tools are now used by caseworkers: a system for extracting and analysing information from interview transcripts, and a system to support decision-making by answering policy-related queries. Norris also noted, in response to a separate enquiry, that the Home Office does not currently use AI software to transcribe asylum interviews — a clarification that distinguishes the new analysis tooling from voice-to-text capture.

Facial age estimation: contract incoming

A separate parliamentary answer from Lord David Hanson confirmed that a £1 million-plus facial age estimation (FAE) contract — originally targeted for February — is now close to award. Hanson told peer Baroness Sally Hamwee that “the FAE procurement outcome was delayed due to a testing and evaluation phase taking longer than originally anticipated”, which has now concluded. Government-led trials previously assessed FAE accuracy across 2.5 million images, with ministers claiming “workable results” cheaply and at speed.

Cross-source context

The asylum AI transparency question is squarely in the territory the ICO is now consulting on. Article 22C of the UK GDPR, in force since 5 February under the Data (Use and Access) Act, requires data controllers to provide meaningful information about automated decision-making and to enable challenge. The ICO’s draft guidance — open for consultation until 29 May — is expected to clarify the threshold for “meaningful information” in the context of partly-automated workflows like the Home Office’s “human-in-the-loop” model. The Hogan Lovells legal commentary published this week explicitly identifies the line between solely-automated and assisted decision-making as the operational pinch point.

Looking forward

For UK public-sector AI deployments more broadly, the Home Office position is a useful test case. If the eventual ICO guidance treats applicant-side disclosure of AI tooling as a baseline requirement under Article 22C, the asylum process will need to update its transparency posture. Until then, claimants challenging adverse decisions will increasingly do so without knowing exactly which AI components informed their case.