AI, Forged Documents and UK Immigration Appeals: Important Lessons from a Recent Tribunal Case

Allegedly forged documents in the light of AI

This case at the Upper Immigration Tribunal, Hoxha & Others 2026, is very interesting and rather complex and – well, depending on your sense of humour – in some ways amusing. 

It was all about allegedly forged documents, and particularly as seen in the modern light of AI; we do not need to tell you that AI has been a hot topic in the immigration world for the last year or two, and of course the light has not always been a good one.

Part of the headnote to the case goes right to the crux of the matter: 

“In light of rapid developments in the use of Artificial Intelligence to generate and manipulate documents and images that are increasingly difficult to distinguish from genuine documents and images…”. 

To put it at its bluntest, some forgeries are so good that nobody can tell for certain whether they are genuine or not. 

This is not a new problem, and it was especially an issue in the case of some documents which had a simple layout and structure and were thus easy to forge. But now, as the Tribunal puts it, AI makes it easier than it was already. 

The Tribunal’s decision relied very heavily on a previous Tribunal decision, Ahmed Tanveer 2002. This case is of course quite old now, and it was heard before the Upper Immigration Tribunal had been invented; in those days it was called the UK Immigration and Asylum Tribunal (UKAIT). This case was “starred”, which means that it was accorded special important and significance. 

The core principles in Ahmed Tanveer have stood the test of time, and they are as follows: 

1. In asylum and human rights cases it is for an individual claimant to show that a document on which he seeks to rely can be relied on.
2. The decision maker should consider whether a document is one on which reliance should properly be placed after looking at all the evidence in the round.
3. Only very rarely will there be the need to make an allegation of forgery, or evidence strong enough to support it. The allegation should not be made without such evidence. Failure to establish the allegation on the balance of probabilities to the higher civil standard does not show that a document is reliable. The decision maker still needs to apply principles 1 and 2.

If you think this looks confusing you are right, and it requires some unpicking. The starting point is that the claimant/appellant must try and show that a document they are submitting is genuine. But a decision-maker will form their own opinion. We know from experience that in some cases the Home Office makes a definite judgement that a document is a forgery but in other cases they do not make such a judgement and the matter is left open.

In such a situation and if the case goes to appeal it will be for the Immigration Judge at the Tribunal to form their opinion. A judge does not have to make a decision on whether a document is a forgery or not; in any case they probably would not have the technical expertise to do so. And, as referred to above, sometimes it is not possible to make such a meaningful decision. 

What the judge does have to do is look at the all the surrounding facts and circumstances and make a rounded decision as to whether the document can be relied on or not – which is not quite the same thing. They might look, for example, at the circumstances in which the document is claimed to have been obtained and assess the level of credibility. In the Ahmed Tanveer case the adjudicator had found that the documents were “unlikely” to be genuine; this was not a finding that they were not genuine but a rather softer finding that they were probably not genuine; the judge looked at the surrounding facts and circumstances and made their decision on this basis. This is deemed to be an acceptable approach.

In the instant case, Hoxha, the facts were quite striking. This case concerned a number of appeal cases held before the First-Tier Tribunal, which shared some specific characteristics. These were all EU Settlement cases (and the court decided that the rules about alleged forged documents now apply to any type of case, not only asylum cases).

These specific characteristics included the following:

• The applicant lodges an application as the family member of an EEA national but not much evidence is submitted.
• The Home Office requests further evidence but it does not come. If an interview is scheduled nobody turns up.
• The application is refused, on the basis of a lack of evidence or non-attendance at interview, and the applicant lodges an appeal with the First-Tier Tribunal on a paper hearing basis (ie a process based entirely on the papers, with no face-to-face interaction). They appeal as an “appellant in person” (ie not represented by a lawyer).
• (And this is verbatim from the court’s decision) “ Often the grounds of appeal and witness statements have the appearance of being generated by a large language model (LLM) generative AI system (overly formal use of language, rigid structuring making heavy use of headings and bullet points, reliance on hallucinations (invented case law) or incorrect citations).”
• Papers for the appeal are submitted very late to the Tribunal and sometimes not submitted to the Home Office at all – and consequently the Home Office does not always get to see them in time or see them at all. (As it is often a paper-based hearing this is more likely to happen, because the judge cannot easily communicate with the Home Office; the whole process is somewhat remote).
• Where the Tribunal has directed that the paper hearing be changed to an oral hearing, the appellant might apply for an adjournment or might withdraw the case or might not turn up for the hearing. 
• (This is also verbatim) “In fact, a striking feature of the appeals is that on closer analysis of the relevant records, it has become apparent that often the same email addresses have been used for several applications by different appellants or sponsors, who are otherwise unconnected.”
• In many cases the appeal is allowed. 

This is a really remarkable pattern, and those of cynical disposition might find something amusing here; there is surely a strong smell of suspicion. 

The issue about paper hearings is an interesting one. Appellants are quite entitled to have a paper hearing rather than an oral hearing if they want to; there is nothing inherently wrong or bad about this. In some cases for example they might not be able to afford the lawyer’s fees for an oral hearing. Another thing is this: sometimes a lawyer can see that an appellant or party would make such a terrible witness that it may not be in the appellant’s best interests to have an oral hearing. In other cases an appellant is simply petrified to appear before a tribunal of law – and who can blame them? But in a situation like this there is a suspicion that a paper hearing is being requested so the appellant can avoid scrutiny from the judge, and this is a rather different kettle of fish. 

Anyway, taking all these egregious facts and circumstances into account, all the allowed appeals were set aside or dismissed and the decision was remade, ie a new negative decision was substituted for the positive one. Perhaps not surprisingly nobody turned up to the Tribunal hearing (and here we are tempted to put a smiley face.)

This decision is very long and complicated and it does not make ideal bedtime reading, but the story it reveals is instructive.

In any event, if you have been accused of providing forged documents we at GSN can try and help you. It is worth bearing in mind that not all cases are as strong as this, and there is a danger that a person who is entirely honest may get tarred with the same brush as those who are not quite so honest. 

Oliver Westmoreland

Senior Immigration Lawyer