ORG calls for Home Office to suspend illegal AI models

In a policy briefing, the Open Rights Group has accused the Home Office of unlawfully using two artificial intelligence (AI) tools to help clear its asylum casework. The Asylum Policy Search tool searches country policy documents, while the Asylum Case Summarisation tool summarises interview transcripts. ORG claims that because both tools generate new text, they risk omitting or filtering crucial facts.

The Home Office confirmed it bypassed the government's AI Playbook, despite both models being launched after the policy's announcement. ORG highlights severe transparency failings: applicants are not informed of AI involvement, cannot correct summaries, and neither tool is featured in the Algorithmic Transparency Recording Standard repository. Furthermore, no data protection impact assessment (DPIA), equality impact assessment (EIA), or standard operating procedure (SOP) has been published for either model, and the staff training provided was deemed inadequate.

ORG has called for the immediate suspension of both models, the publication of impact assessments and the registration of the tools on the ATRS repository, along with independent investigations by the office of the Chief Inspector of Borders and Immigration.

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.