A senior judge said the Home Office used an AI-hallucinated report to reject an asylum claim. The report doesn't exist.

A senior immigration judge has found that the Home Office refused a Moroccan woman's asylum claim by citing a country-safety report that appears never to have existed. The case, [UK and R (on the application of Munir) v Secretary of State for the Home Department [2026] UKUT 00081 (IAC)](https://tribunalsdecisions.service.gov.uk/utiac/2026-ukut-00081), was handed down on 14 July. The claimant, a Moroccan woman with her child, told the tribunal she had fled forced underage marriage, rape, and extreme violence at the hands of a previously convicted husband. The Home Office refused her claim. A first-tier tribunal (the standard immigration court) upheld the refusal on the same basis. On further appeal, the Upper Tribunal (the appeal court) was told no such report had ever been published.

"It appears that no such CPIN has ever existed," the judge wrote. CPIN stands for country policy information note, an internal Home Office report that immigration judges normally treat as authoritative. The judge also noted that the Home Office's own country policy and information team had been unable to confirm the document's existence. After the Guardian asked for the document, the department produced a different, unpublished internal note (a country information note, or CIN, a separate, less formal guidance document) that had never been disclosed in the proceedings and did not support the refusal. No country policy and information note on Morocco has been published since 2017.

The judge's AI framing is provisional. The refusal letter "bears hallmarks consistent with the use of artificial intelligence" and the cited report "may" be an AI "hallucination", a confident but fabricated output from a generative model. The judge called the issue a "procedural irregularity" rather than a substantive finding on the merits. The classification leaves the underlying asylum claim alive but means the AI-hallucination explanation is, on the record, the judge's working hypothesis, not a proven fact.

The ruling exposes the verification gap: nothing in the Home Office's process caught that the cited document was fictitious. The Home Office's own published evaluation of its asylum AI trials describes efficiency gains without documenting a verification layer for cited sources. Open Rights Group and Gatehouse Law have raised the same concern: a model that drafts a refusal letter faster is also a model that invents a citation faster.

Munir sits in a line of Upper Tribunal cases that the practitioner side of UK immigration law knows as the "Hamid" warnings, named for an earlier ruling that flagged fabricated citations in representatives' filings. The Hamid line was about lawyers misusing AI. Munir shifts the AI-citation scandal in UK immigration from representatives to the decision-maker, with the same guardrail problem. Executive Immigration News, the legal trade outlet tracking these rulings, and Damien Charlotin's running "Hallucinations" log of fabricated citations in court filings both treat Munir as an escalation on the misuse axis, not a novelty.

The judge classified the issue as a procedural irregularity rather than merits error, which means the ruling does not impose a general verification duty on the Home Office for future refusals. That duty would have to come from a higher court, from legislation, or from a settlement the department accepts in the underlying case.

Three moves will show whether Munir changes anything beyond the case itself: the Home Office's formal response to the judgment, any audit it announces of prior cases that cited the same note, and whether the upper tribunal's published decision circulates into a disclosure safeguard for future asylum refusals.