Home Office AI trial found serious errors in 9 per cent of asylum cases

The Home Office is rolling out a new artificial intelligence (AI) tool to address the backlog of asylum claims, despite it making serious errors during a trial. According to government documents, 9 per cent of the summaries produced by the Asylum Case Summarisation (ACS) were so flawed that they were removed from a pilot.

Immigration Minister Angela Eagle praised the ACS tool intended to speed up application processing times as a success that could cut summarisation time by a third. However, an evaluation also revealed that 23% of caseworkers lacked full confidence in the tool's output.

MPs and refugee charities have voiced concerns that thousands of people, including those with "life and death" cases, could be affected. Labour MP Tony Vaughan KC supports the use of AI technology but urges the government to improve the tool, insisting that caseworkers should continue to assess entire transcripts, not just the AI summaries, until the error rate is fixed. The ACS tool is expected to be fully rolled out by January 2026.

In a statement, Jake Hurfurt, head of research and investigations at Big Brother Watch, said: "Outsourcing vital decisions to error-riddled AI tools is a worrying path for the government to go down. The government should stop seeking to cut costs with shiny AI tools and instead properly invest in staff."

Training Announcement: The BCS Foundation Certificate in AI examines the challenges and risks associated with AI projects, such as those related to privacy, transparency and potential biases in algorithms that could lead to unintended consequences. Explore the role of data, effective risk management strategies, compliance requirements, and ongoing governance of the AI lifecycle and become a certified AI Governance professional. Find out more.