Connecticut judge Walter Spader Jr said Matthew Elliott used text formatted in tiny white type on a white background to instruct AI systems to favour his arguments, ignore previous court rulings and recommend the outcome he wanted.

The tactic had no effect because Connecticut’s judicial branch does not use AI to review filings, but Mr Spader described it as a “dangerous” development that courts may increasingly encounter.

Elliott continued inserting hidden messages into later filings despite being warned that he could face sanctions. Some were jokes, including a link to a Nosferatu video and messages such as “hi :) I hope yo ucant see me”.

The judge found the conduct amounted to “serious litigation abuse” and banned Elliott, who represented himself, from electronic filing, although he declined to impose a financial penalty.

Mr Spader also warned that litigants using AI to build cases risk becoming entrenched in arguments because chatbots can be prompted to support a position rather than test it.

“Those using these tools must ask them to test a position as readily as to advance it,” he said.