Open Rights Group raises concerns over Home Office AI asylum tools and calls for their suspension

A new report by the Open Rights Group (ORG) has raised concerns about the Home Office's use of artificial intelligence (AI) in asylum decisions, with a six-month investigation highlighting significant error in content generated by AI tools and a lack of transparency and independent oversight.

The 20-page report can be downloaded here.

The Open Rights Group examined two AI tools used by the Home Office: the Asylum Case Summarisation (ACS), tool which summarises asylum interview transcripts, and the Asylum Policy Search (APS) tools, which helps caseworkers search country policy information. Both use OpenAI's GPT-4o model through Microsoft's Azure AI Foundry.

As the report highlights, both AI tools generate their own new text rather than merely retrieving existing text and they can filter or omit facts relevant to a decision.

A 2025 Home Office pilot evaluation found that 9% of ACS summaries were inaccurate or omitted information and had to be removed from the trial. However, subsequent sampling by Subject Matter Experts (SMEs) of 203 summaries after the tool's wider rollout found that 22% failed an initial assessment and 30 of 56 summaries examined at a second stage also failed.

Whilst noting that the exact SME procedures to test for a 'fail' were unclear, the Open Rights Group noted: "Therefore, across both stage reviews 22% and 15% of sampled summaries were assessed as not fit for use – significantly higher than the 9% figure in the Home Office's published pilot evaluation."

The Open Rights Group further notes that only 203 summaries were checked by Home Office SMEs between April and July 2026, representing only approximately 0.3% of the estimated 70,000 cases processed during that period.

According to the report, the situation for the APS tool is worse and there is no equivalent ongoing quality-control process. Anna Turley, the Minister of State for Border Security and Asylum, confirmed in August that the testing of the APS tool had now ceased.

The report further criticised the Home Office for not applying the UK Government's AI Playbook to the tools, which offers guidance on using AI safely, effectively and securely. The Home Office is also not planning an independent audit of its AI tools and is not listing them on the Algorithmic Transparency Recording Standard Hub.

Responses to parliamentary questions and freedom of information (FOI) requests contained contradictory accounts of staff training and testing of the AI tools, the Open Rights Group said. Some FOI requests were refused.

The report states: "The Home Office has refused to provide key information including about staff training, the prompts being used, and how the tools are being tested for quality assurance. Their replies have been at times self-contradictory. Our findings suggest that these tools have been rolled out without proper testing, quality assurance and training. This careless approach could have catastrophic consequences for those people whose future is in the hands of Home Office decision-making."

In addition, the Open Rights Group questioned whether the tools had delivered the efficiency improvements claimed by ministers. Its analysis of Home Office data found that the average monthly number of initial substantive asylum decisions since April 2026 had fallen by about 42% compared with the same period in 2025, while the mean time between substantive interview and decision had increased by 29%.

The report calls on the Home Office to suspend both tools, commission an independent evaluation and audit, and ask the Independent Chief Inspector of Borders and Immigration to investigate their use.

It says there should be no use of AI in the asylum decision-making process without proper consultation, adherence to domestic regulations and international obligations, and full alignment with human rights principles