Law360, London (May 5, 2026, 4:10 PM BST) -- Justice Colin Birss said Tuesday that he is improving his judgments by using artificial intelligence to check for clarity and consistency — but he hinted that having the tech write rulings from scratch would be a step too far.

The Court of Appeal justice said that, once he has finished writing a judgment, he uploads the document into the judiciary's secure AI chatbot. The system then checks for any inconsistencies within the ruling — and while the AI tool is not always correct, it often spots chances to improve clarity, Justice Birss said at an International Trademark Association event in London.

"That, to my mind, is an extremely helpful use of AI," Justice Birss added. "It makes me produce better judgments. It's not judging the case, but it is helping me."

Justice Birss noted that judicial clerks and assistants can do a similar job. But the "reality" is that AI tools can check judgments in a matter of seconds, he said.

But he warned that judges ultimately take full personal responsibility for the rulings bearing their names. The justice said he's unaware of any judges using AI to write judgments entirely.

"The idea that a judge could put out a judgment and then say, 'oh, I'm very sorry, I've cited a case that doesn't exist. It was the AI's fault' … It's obviously ridiculous," Justice Birss said.

The Court of Appeal justice was speaking on a panel that also included High Court Judge James Mellor, who ruled in 2024 that Australian computer scientist Craig Wright was not Satoshi Nakamoto, the pseudonymous inventor of bitcoin.

Judge Mellor cited Wright's case to highlight the risks of litigants using AI to help draft court filings. He said Wright had "clearly" used AI to draft a 700-page witness statement a week before the High Court trial. The Australian then filed 900 pages of grounds for appeal that contained many hallucinated cases, Judge Mellor said.

The High Court judge added that AI can produce "what looks like a very convincing document, but when you know the area, you realize it has invented a whole bunch of cases."

But in some instances, AI-generated filings are partially correct and may even be "quite cogently written," Justice Birss said. "There's an access to justice issue. You can't just tell people they can't use these tools."

High Court Judge Richard Meade was also on the panel. He told the audience that he also believes AI can be "quite positive in some cases" when people are representing themselves. For example, litigants in person can use the tech to help fill in court forms that might otherwise be "quite daunting," Judge Meade said.

--Editing by Joe Millis.

For a reprint of this article, please contact reprints@law360.com.