If you’ve been following the legal world in 2025, you know that artificial intelligence is no longer just a buzzword tossed around at tech conferences — it’s a daily reality for lawyers, judges, and even the unlucky souls who end up in court. But let’s be honest: the journey hasn’t exactly been smooth sailing.
AI Crashes the Courthouse
AI first made its way into law firms as a handy research assistant, promising to cut down on endless hours spent digging through case law and drafting briefs. Take, for example, e-discovery platforms that could sift through mountains of emails faster than any paralegal, or contract analysis software that flagged risky clauses in seconds.
By now, the majority of lawyers have tried out AI tools for legal work, and some are using them regularly. Last month, Thomson Reuters published the results of its annual Future of Professionals Report, which surveyed a wide range of attorneys. The survey found that among legal professionals currently using AI tools, 77% use them for document review, 74% use them for legal research, 74% use them to summarize documents, and 59% use them to draft briefs or memos.
Over half of respondents said their organizations are already seeing an ROI from investing in AI. The hope was that these digital helpers would make life easier, freeing up time for actual lawyering. And it’s certainly done that.
When Tech Doesn’t Deliver
But 2025 has shown us that trusting AI isn’t as simple as clicking “generate.” Over the past couple of years, courts across the country have had to deal with a new headache: “hallucinations.” That’s when AI confidently spits out fake case citations or invents legal arguments that sound real but don’t exist. Legal educators are sounding the alarm about “AI literacy”—knowing both the strengths and pitfalls of these tools.
It started with smaller firms and solo practitioners getting sanctioned for filing briefs full of made-up cases. But now even big names like Morgan & Morgan and Butler Snow have landed in hot water after submitting AI-generated fictions to federal judges. And these judges have made it clear: no matter how advanced your tech is, lawyers are still responsible for vetting every word that goes into a court filing. Some law firms are scrambling to review thousands of citations and update their policies on AI use.
As mistakes made by AI keep popping up in legal documents, some judges are trying out new ways to stop lawyers from cheating with chatbots (at least without double-checking their work). Instead of just handing out fines, judges are getting creative with their punishments.
A Judge's Novel Punishment
Take what happened in Nevada. Two defense attorneys from Cozen O’Connor admitted they filed paperwork with AI-generated content. Judge David Hardy gave them a choice: either pay $2,500 each, get kicked off the case, and be reported to the state bar, or write letters to their old law school deans and bar officials explaining what went wrong and offer to give talks about AI and ethics.
Cozen O’Connor says it has strict rules about using AI. It apparently has a policy against using public tools for client work, along with mandatory tech training and a requirement that lawyers check everything themselves. They stressed that breaking these rules leads to consequences. But despite these purported rules, lawyers will get lazy and sloppy.
The case at issue with Cozen O’Connor was about internet provider Uprise and a failed fiber optic project. Judge Hardy found at least 14 citations that looked made up, plus others that were misquoted or twisted. The lawyers and their firm apologized and admitted that one of them had used ChatGPT to help write the document…and accidentally submitted an early draft. (To no one’s surprise, he’s since been let go from the firm.)
Creative Consequences on the Rise
Judge Hardy isn’t alone in this approach. Gary Marchant from Arizona State University notes that judges are getting more frustrated as these problems keep happening. That’s why some are looking for new ways to not just punish sloppy lawyers but also send a message to everyone else.
Some judges have turned to making lawyers who abuse AI apologize or at least warn affected judges. A federal judge in Arizona required a lawyer to notify other judges who had been falsely cited as authors of non-existent cases generated by AI. The judge explained that alerting her counterparts would "minimize potential harms of those fictitious cases to those jurists." The lawyer was also removed from the case.
In New York, Judge P. Kevin Castel ordered the attorneys of Levidow, Levidow & Oberman to write letters to each judge falsely identified as the author of one of various fake cases, informing them of the error and apologizing. In Alabama, Judge Anna Manasco ordered Butler Snow attorneys to share her ruling with clients, other judges and lawyers at their firm nationwide. She argued that making up legal authority calls for much more serious consequences than just the usual warnings or small fines.
Will lawyers shape up, or will judges keep having to come up with weirder sanctions? Either way, the moral of the story is the same one you’ve been hearing since write-on for Law Review: cite-checking is important!
Related Resources:
- Does the Slowdown of AI Adoption by Large Companies Indicate Changing Attitudes? (FindLaw's Practice of Law)
- Haven't Started Using AI Yet? You Aren't as Behind as You Might Think. (FindLaw's Practice of Law)
- New Bill Proposes To Ease Regulations for AI Tech Companies (FindLaw's Law and Daily Life)
