AI asylum seeker: UK to trial algorithm for asylum triage

The UK government is preparing to trial an AI-assisted triage system for asylum claims, intended to help caseworkers prioritise applications and reduce backlogs. As indicated by available reports, the Home Office claims that the tool will not make final decisions, but critics argue that even “support” systems can shape outcomes through risk scoring and routing. The pilot is expected to begin in late 2026, according to departmental planning documents.

The project follows earlier experimentation with data-driven screening in immigration processes. Officials have indicated the trial will focus on routing cases to different queues rather than issuing approvals or refusals. In a middle stage of the pilot, the system will be tested on a limited cohort of new claims received at Lunar House in Croydon. Campaigners say a Home Office tool used upstream could still influence what evidence is requested and how quickly applicants are interviewed. Ministers have described the approach as “modernisation,” while lawyers warn it may embed assumptions about credibility.

Transparency is likely to face questions since the government has not published the model’s full feature set, nor clarified how it will be audited for disparate impact. A central concern is that an AI asylum seeker triage score could correlate with nationality, travel route, or language, even if those fields are not explicitly weighted. In April 2026, MPs on the Home Affairs Select Committee requested clearer documentation on how automated tools are tested before deployment. Civil society groups have called for publication of evaluation metrics and error rates.

Oversight will also matter, particularly given the increasing role of contractors in public-sector technology delivery. The Home Office has said the pilot will be monitored by internal governance processes and may be reviewed by the Information Commissioner’s Office in relation to data protection compliance. In 2025, the department reported a sharp increase in asylum casework staffing, which it says should reduce pressure to rely on automation. However, advocates argue that accountability frameworks must be clear before any triage model is scaled.

Legal experts note that even when a tool is formally “advisory,” the practical reality may be that staff follow system recommendations, especially under time pressure. The government says caseworkers will retain discretion and that applicants can challenge decisions through existing routes, with guidance expected to be updated alongside the pilot. Human rights organisations argue that, if errors occur, it may be difficult for individuals to understand or contest a model-driven routing choice without meaningful explanations. Further details may emerge through parliamentary questions and future disclosures under the Freedom of Information Act 2000.