It's easy to understand the temptation. Instead of spending time and money poring through caselaw to find support and precedent for an argument in a filing for a client, a well-crafted request to generative artificial intelligence (GAI) can return a bevy of responses in a flash. Not only do they fit your stance perfectly, it would have taken you forever to find some of the more obscure ones.
Unfortunately, finding that elusive case without GAI might have taken longer than forever, because it might not actually exist. The unreliability of generative AI was proven yet again in the recent Mike Lindell defamation lawsuit. U.S. District Judge Nina Wang fined Mike Lindell's attorneys $3,000 each for submitting a filing created by GAI that contained over two dozen mistakes. These included so-called GAI "hallucinations" of examples drawn from cases that don't exist.
While it seems likely that GAI is destined to have a presence in courtrooms, the type and extent of that presence are still being hashed out. With GAI-generated errors becoming more common in legal filings, the looming question is how much the tool can be trusted.
Is a Craftsman Only as Good as Their Tools?
Artificial intelligence (AI) is the ability of a computer or computer system to mimic human cognitive abilities for problem-solving, decision-making, and learning. GAI takes it a step further by generating content from what it has learned from other sources, known as scraping. The legality of this process is the source of more than a handful of current legal battles being waged in courtrooms.
One of the issues discovered with large language models (LLMs) like GAI is that they are prone to creating "hallucinations." These can appear to be plausible information but are actually fabricated, inaccurate, or misleading. GAI hallucinations occur because the LLM is trained to predict a probable sequence or pattern based on the data it was fed.
As you might imagine, this proves extremely problematic when a prompt elicits results that include citations of legal cases that don't actually exist. It's further compounded by attorneys who don't fact-check the GAI's work before submitting it. The American Bar Association (ABA) felt that these issues, combined with required adjustments in client billing due to GAI, required a formal opinion in 2024.
GAI has the potential to be an extremely useful tool for legal research and the preliminary drafting of legal documents, but needs to be properly harnessed. Increased use of GAI by legal professionals has resulted in an explosion of cases with hallucinations and other filing mistakes. A site that tracks legal proceedings marred by GAI issues has gone from recording 11 in 2023 to well over 125 examples in the first six months of 2025.
That's the Guy With the Pillows, Right?
A recent example of GAI misuse occurred in the high-profile defamation case involving MyPillow creator Mike Lindell and Eric Coomer, a former employee of Dominion Voting Systems. In his numerous unsubstantiated claims of voting fraud in the 2020 elections, Lindell frequently singled out Coomer as responsible for the alleged crimes during radio and television appearances.
Coomer sued Lindell and other defendants for defamation, intentional infliction of emotional distress, and civil conspiracy. He claimed that Lindell and others' constant harassment forced him to exit his career. In June 2025, a jury awarded him approximately $2.3 million in damages. That wasn't the end of the legal drama.
On July 7, 2025, Judge Wang issued Christopher Kachouroff and Jennifer DeMaster, Lindell's lawyers in the case, fines of $3,000 each. The sanctions resulted from a GAI-created filing submitted in February using a combination of Co-Pilot, Westlaw’s AI, Gemini, Grok, Claude, ChatGPT, and Perplexity.
According to Judge Wang, the filing contained over two dozen mistakes, including citations of non-existent cases. Kachouroff claimed a draft copy had been filed by mistake, but while assessing the fines, Judge Wang accused the pair of not being forthcoming about using GAI until faced with incontrovertible proof.
There's no doubt that GAI can serve as a powerful tool for legal professionals and simplify a number of tasks. GAI has been discovered creating false cases, attributing fake quotes to existing cases, and citing actual cases with citations that don't support the argument correctly. How useful a tool it will ultimately be is likely to be part of a long and hallucination-filled process.
Related Resources
- What Are Legal Ethics and Professional Responsibility? (FindLaw's Choosing a Lawyer)
- Pro Se Litigant Fined 10K for Filing AI-Generated Reply Brief (FindLaw's Practice of Law)
- Attorney Faces Sanctions for Citing Fake Cases From ChatGPT (FindLaw's Greedy Associates)
