According to The Guardian, a Moroccan woman and her child sought asylum after fleeing forced child marriage, rape, and death threats from her husband. The Home Office rejected the claim by citing a country guidance document supposedly proving Morocco was safe for her return. When the tribunal tried to locate the document, it didn't exist. A senior judge raised the possibility that AI hallucination had fabricated the citation.
This isn't a filing error. In asylum cases, a false citation can send someone back to the person or place they're fleeing. Once a tribunal repeats the error, invented facts calcify into state precedent. The Home Office's use of AI in immigration decisions now carries documented risk of life-threatening consequences.
The immediate test: can the Home Office prove where this document originated and how it entered the case file? The harder question follows: will ministers and courts treat this as an isolated embarrassment, or as evidence that AI use in asylum adjudication requires real audit trails and human verification before any decision lands.
Governments love AI for speed and volume. But asylum law is the wrong place for synthetic confidence. If officials can't show the source for a claim about country safety, they're not making a judgment. They're gambling with a woman and a child under the seal of the state.
Filed to the World desk · Jul 28