The Aftermath Of An AI Trip To Court Over A Block Transfer Order 

There are many things a judge expects to see in block transfer order proceedings.

A reworded version of an insolvency rule generated by what appears to be an AI assistant is not usually one of them.

Yet that appears to have unfolded in Anthony Malcolm Cork & Anor v Mark Smith [2026] EWHC 1199 (Ch), a judgment that could become a regular flashback for law firm compliance departments, as a potential flurry of reviews and perhaps even redrafted AI policies could sprout.

This has become a most unfortunate case in what are notably not contentious proceedings. The case itself concerned one of those technical insolvency procedures known as a “block transfer application”. Normally, these applications are “administrative and uncontentious”, handled as part of a judge’s daily diet of boxwork.

Then came the AI hallucination.

The then Applicants’ solicitors, in a letter to the court, asserted that the court had a power concerning liquidator release to be found in “Rule 12.37(5)” of the Insolvency Rules 2016, before quoting a drafted provision which, unfortunately, did not exist.

The relevant AI hallucinogenic-derived rule read:

“The court may, in an order made under this rule, make such provision as it thinks fit with respect to matters arising in connection with the replacement, including— (a) the release of the outgoing officeholder…”

Judge Mullen’s reaction was:

“I therefore checked IR 12.37(5) on the legislation.gov.uk website and in a practitioner text to find that it says nothing of the sort.”

And then comes the line about the hallucinogenic properties that may sometimes be attributed to AI:

“Inevitably, my first thought was that this was an artificial intelligence (“AI”) “hallucination”…”

The judgment explains what was meant by those hallucinogenic properties:

““Hallucination” is a term used where AI generates false or misleading information, which it presents as fact.” 

The court was not especially persuaded by the follow-up explanation provided as to what had transpired.

The key issue was set out as follows:

“The Purported Text … struck me as likely to be an AI hallucination, which had not been checked.”

People may now understand that large language models are essentially probability engines with conceivably impeccable grammar handed down from AI celestial cities, but with a potentially imperfect relationship with reality.

The judge said you:

“… cannot outsource the process of legal research or of legal reasoning to an AI. It is a tool to be used with caution. AI has the potential to be wholly unreliable.”