By now, AI’s public blunders are legion, and many of the most jaw-dropping have come from one field: law. But that’s not because lawyers are sloppier than other workers; it’s because a judge is the one looking at the work product. Brad Harmon of HunterMaclean offers a sobering angle: The types of headline-making AI hallucinations playing out in law are happening elsewhere, too, you just don’t know about it yet.
Here’s something most companies don’t want to talk about: their people are already using AI at work, and leadership has no idea how. No policy, little vetting, thin guardrails. Employees are uploading sensitive documents to public models, submitting AI-generated work without review and making decisions based on outputs that are often unverified. It’s happening in marketing departments, finance teams, consulting firms and HR offices. The only reason it hasn’t blown up yet is that the consequences are usually invisible.
Nobody gets publicly sanctioned for a hallucinated slide deck during a lunch-and-learn. In my profession, they do.
I’ve been a trial lawyer for over 20 years, and I’ve managed a 60-attorney firm. Law has become, almost by accident, the first profession where unmanaged AI adoption has produced visible, documented and very public consequences. Attorneys have been sanctioned for submitting fabricated case law generated by AI. Cases have been thrown out. Careers have been irreparably damaged. And it’s all on the record.
That makes law a useful case study, not because lawyers are uniquely reckless, but because our mistakes happen in public. The same patterns playing out in every other profession just don’t have a judge waiting at the end to check the work.
The numbers should make any leader uncomfortable, regardless of industry. Microsoft says that 75% of knowledge workers now use AI at work, with usage nearly doubling in just six months. And 78% are bringing their own tools to work without employer guidance or clearance. More than half (57%) are reluctant to tell their manager they’re doing it, not out of embarrassment or fear of job loss, but because nobody has trained them.
Your people are already using these tools, without permission and without guardrails, and most of them aren’t saying a word.
AI: Reliable or Reliably Unsafe?
Recent lawsuits over AI applicant-screening tools highlight important differences
Read moreDetailsWhen AI goes wrong, it goes wrong in public
In law, we’ve all seen what unsupervised adoption can do. A California attorney was fined $10,000 after 21 of 23 case citations in his appellate brief turned out to be fabricated. He told the court he hadn’t read the AI-generated text before filing it. In Colorado, an attorney accepted a 90-day suspension after acknowledging in a text message that he hadn’t checked an AI-drafted motion for fabrications. A researcher who tracks AI sanctions across US courts has documented more than 1,000 cases since 2023.
The pattern in every case is the same: Someone used the tool, skipped the verification and nobody at the firm had written a policy saying they had to. Swap “brief” for “proposal,” “investor deck” or “compliance report” and the exposure is identical.
The irony is that AI, used well, has genuinely revolutionized the way I practice law. That’s something many veteran attorneys are hesitant to admit publicly. I can feed a 200-page deposition transcript into a language model and ask it to identify every statement a witness made about a specific topic, organized chronologically with page and line citations. That used to take an associate an entire afternoon. A well-structured prompt can produce a first draft of a legal motion that hits 70% to 80% of the arguments I’ve already identified, organized logically with the right legal standards. I spend my time on the remainder that requires judgment, strategy and knowledge that only comes from years of practice.
I also find AI is remarkably good at pressure-testing my thinking. I upload my work and ask it to identify the strongest counterarguments the other side is likely to raise. I ask it to find the weaknesses in my reasoning. That’s a workflow any strategist, analyst or executive could use tomorrow to improve their output.
But none of that value matters if people are using these tools without oversight. And right now, in most organizations, that’s exactly what’s happening.
AI still hallucinates. It fabricates citations, invents data points and presents nonsense with total confidence. In law, the verification failure is obvious — a fake case gets flagged by a judge. In other fields, a hallucinated statistic lands in a board presentation, a fabricated source makes it into a published report or a confidential dataset gets fed into a model that trains on user inputs. The damage is just as real but harder to trace.
Regulators are starting to respond, and the legal profession is being forced to figure out governance on an accelerated timeline. And when the nature of law firms is change-averse, that can be hard. Regardless of the outcome in law, every other industry should assume they’re next.
The fix is simpler than you think
You don’t need a six-figure technology budget. You don’t need a dedicated innovation team. You need written guidance, two or three vetted tools and employees who own the initiative to use them. You need to decide what can be uploaded to external models and what can’t. You need a verification standard for AI-generated work product. You need people trained not just on how to use these tools but when and how not to.
AI won’t replace professionals who do complex, judgment-intensive work. But it will push out the ones who don’t do it efficiently, and the organizations that sort out governance first will capture the value without absorbing the risk.
The competitive window has opened. But it won’t stay open forever. And right now, while leadership debates whether to even have the AI conversation, their people have already made the decision for them.
Brad Harmon is a litigator and former managing partner at HunterMaclean in Savannah, Ga., where he brings over 20 years of experience practicing complex commercial litigation and healthcare law. 
