Judge Sandy Kemp insisted that he did not us AI to help him write his 312-page judgment in the case, which has now been corrected a number of times.

His claim followed a formal complaint of judicial misconduct made by retired lawyer and former part-time tribunal chair Ewan Kennedy, who said that the corrections went far beyond those permitted under section 67 of the Employment Tribunal Rules 2024, which provides for the correction of clerical mistakes.

He said it was “a very serious matter” for corrections to stray into “substantial matters”, among them substituting a word with its opposite.

Mr Kennedy said: “Suggestions have been made in the press that the falsehoods may have resulted from careless use of some form of artificial intelligence, but that can be no possible excuse.

“Any solicitor who did something similar in professional practice could expect immediate disqualification for life.”

Sandra Muir, speaking on behalf of Judge Susan Walker, the president of Employment Tribunals, said that the mistakes in the judgment and the “inclusion of misquotations from other judgments” were “regrettable”.

She said she was “aware of the widespread public speculation that Judge Kemp has improperly used generative AI in the writing of this judgment”.

Ms Muir added: “Having made enquiries, I am satisfied that Judge Kemp did not use generative AI in drafting the judgment.

“It is clear from my enquiries that the source of the erroneous quotes from Forstater and Ashers was an exchange of correspondence between Judge Kemp and a judicial colleague.”

Ms Muir added, however, that Judge Kemp accepted he was “ultimately responsible” for the content of the judgment and had taken steps to fix the errors “promptly” after they were identified.