The case centers on a Moroccan woman and her child, and on whether a government refusal letter cited evidence that may never have existed. It raises a sharper question for public bodies: what happens when AI-sounding errors enter life-changing decisions?
A judge suggested the Home Office used bogus AI-generated evidence to refuse an asylum claim in the United Kingdom, after a tribunal case involving a Moroccan woman and her child turned on a Morocco document that could not be found. The court concern matters because the missing evidence helped justify saying Morocco was safe for her.
The woman says she fled forced underage marriage and extreme violence, and feared being killed by her husband. A senior immigration judge has now flagged the Home Office refusal letter as bearing hallmarks of AI use, according to the decision reported by the Guardian.
A missing Morocco document
At the center of the dispute is a document described as a country policy information note, or CPIN. These Home Office documents are used in asylum cases to summarize country conditions and can carry real weight when officials and judges assess whether someone can safely be returned.
In this case, the Home Office refused the woman’s claim after referring to a supposed Morocco CPIN from July 2021. The refusal decision treated that material as authoritative evidence that Morocco would be safe for her.
The problem, according to the upper tribunal judge, is that the document has not been located. In a decision dated 14 July, the judge said it appeared that no such CPIN had ever existed, and that the Home Office’s own country policy and information team had been unable to confirm it.
That is why the case has moved beyond a narrow paperwork dispute. If a non-existent document was used to support refusal of protection, the error goes directly to the fairness of the asylum process.
Why one acronym mattered
The case turns partly on the difference between two similar-sounding terms: CPIN and CIN. A CPIN is generally a public country policy information note. A CIN is an internal country information note.
According to the Guardian’s report, when asked about the missing document, the Home Office initially pointed to an archive where the cited CPIN could not be found. It later produced a different document: an unpublished internal CIN.
That distinction matters because the unpublished internal note was not disclosed to the court during the legal proceedings. It also did not appear to support the broad claim that Morocco would be safe for the woman in the way the refusal letter suggested.
The Guardian reported that no CPIN on Morocco has been published since 2017. During the case, references appear to have shifted from internal guidance to a supposedly public policy document, a change that could have affected how much confidence the tribunal placed in the evidence.
The judge’s AI concern
The upper tribunal judge did not make a final finding that AI was used. He did, however, say the Home Office refusal letter bore features consistent with artificial intelligence and that the reference to the July 2021 Morocco note may have been the result of an AI hallucination.
AI hallucination is the term used when a system generates information that sounds plausible but is false. In a consumer chatbot, that can be annoying. In an asylum decision, it can shape whether a person is granted protection or told they can be removed.
The judge described reliance on a document that did not exist as analogous to relying on bogus evidence. He also found the issue amounted to a procedural irregularity.
That wording is significant. It frames the alleged problem not as a harmless drafting slip, but as a potential failure in the process used to decide a claim involving personal safety.
The human stakes behind evidence
The unnamed woman’s claim is not abstract. She said she fled Morocco after forced underage marriage and severe violence, including rape, and feared her husband would kill her if she returned. Her husband was described in the case as powerful and previously convicted.
Country evidence is often central in asylum cases like this because decision-makers must assess both the individual story and the conditions in the country of origin. If the country evidence is wrong, missing or misrepresented, the entire risk assessment can be distorted.
The internal Morocco note later produced by the Home Office, according to the Guardian, included details that complicate any simple claim of safety. It referred to 40,000 exemptions granted for child marriages in 2018, one-fifth of marriages that year.
It also said Morocco’s laws do not explicitly criminalize marital rape and that women who report it may face criminalization for sex outside marriage if authorities disbelieve them. Those details cut against a neat conclusion that a woman alleging this kind of abuse would necessarily be protected.
The Home Office response
Home Office sources told the Guardian they were able to provide additional information requested by the judge. A department spokesperson gave a broader statement on asylum policy, saying the home secretary was restoring order and control to the system and increasing removals of people with no right to remain in the UK.
The spokesperson also said the asylum system is kept under constant review and that claims are considered on their individual merits using the latest evidence, so that protection goes to those in genuine need.
That response does not directly answer every question raised by the tribunal. The unresolved issues include how the CPIN reference entered the refusal letter, whether any AI tool was involved, and what safeguards exist to prevent non-existent sources from being cited in official immigration decisions.
There is also a narrower possibility: the error may have come from confusion between an internal CIN and a public CPIN rather than from AI. But the judge’s concern shows why that distinction may not be enough. Either way, the claimant and the tribunal were left dealing with evidence that could not be verified in the form cited.
What this case could change
Governments are under pressure to process immigration and asylum cases quickly. Digital tools can help with administration, translation, document sorting and research. The danger comes when speed weakens accountability, especially if decision letters cite sources that lawyers, judges and claimants cannot inspect.
For public bodies, the lesson is not simply that AI is risky. It is that any automated or AI-assisted drafting must leave a reliable audit trail. If a refusal letter cites a document, that document needs to exist, be identifiable and be available to the parties who may have to challenge it.
The woman’s case continues, so the tribunal has not produced the final answer on her asylum claim. The immediate damage, however, is already clear: trust in the evidence behind a life-changing decision has been shaken.
If the judge’s suspicion proves correct, the case may become an early warning about AI hallucinations in government decision-making. If it proves to be a human drafting or disclosure failure, the core problem remains much the same: asylum decisions cannot rest on evidence no one can find.
Leave a Reply