A recent milestone in AI was the announcement that LLMs could pass the bar exam. I was impressed as someone who knows nothing about the bar beyond its gruelling reputation in Hollywood movies.

But I do understand that the legal industry thrives on nuance. As a tech nerd, I know nuance and code don't always play nicely together, which has landed some lawyers in very hot water.

What happened?

Lawyers were found to be submitting court filings with made-up case citations, eventually getting slapped by judges and paying massive fines. All because they trusted ChatGPT like it was Google instead of what it is—a confident fiction machine.

This isn't about lawyers being dumb—they're some of the most time-poor professionals—it's about the fundamental mistake everyone's making with AI.

The "Super Search Engine" Delusion

One sanctioned attorney thought ChatGPT was a "super search engine." Here's the problem: It's a pattern-matching text predictor that sounds confident about things that don't exist.

This delusion is everywhere. People think AI tools are databases. They're not. They're sophisticated autocomplete systems that sometimes nail it and sometimes fabricate entire realities with the same confident tone.

The price of this delusion? In Mata v. Avianca, lawyers got hit with $5,000 in sanctions for submitting six entirely fictional cases. The latest penalty? $31,100 against two major law firms for fake citations in a State Farm case.

The Real Lesson: Human + AI > AI Alone

Here's what the competent lawyers figured out: Treat AI like a junior associate, not a senior partner.

Alexander Kolodin, an Arizona election lawyer, nailed it: "You don't just typically send out a junior associate's work product without checking the citations." He should know—he openly used ChatGPT to write Arizona's deepfake law, which passed unanimously 57-0.

The survival strategy isn't avoiding AI—it's understanding what it does.

Even AI companies aren't immune. Anthropic's lawyers used Claude AI to generate a fake citation while defending the company against copyright claims. The irony was brutal—getting caught using your client's AI to fabricate sources.

AI Guardrails for Lawyers

The lawyers who aren't getting embarrassed in court follow three simple rules:

1. Never trust; always verify. Judge Kathryn Kimball Mizelle found nine AI hallucinations in a single filing and struck the entire motion. Human eyes check every citation, fact, and confident-sounding claim.

2. Understand what you're working with. AI isn't a search engine; it's a writing assistant with a problem with confidence. Even Stanford misinformation expert Jeff Hancock got burned when ChatGPT fabricated academic citations in his court testimony.

3. Use AI for what it's good at—first drafts, brainstorming, and organizing information—not final answers.

4. If you're paying for legal documentation (or any documentation), run it through ZeroGPT or Quillbot to check out the AI-generated portion. With these tools, exercise caution as the results can vary (I use Grammarly to tidy my writing, which gets detected by some AI detection tools, so use these with discretion).

The Bigger Picture

30.2% of attorneys now use AI tools, up from 11% in 2023. The American Bar Association issued its first formal AI guidance in July 2024. Legal databases like LexisNexis and Westlaw are integrating AI everywhere. This train isn't stopping.

Dean of Suffolk Law School, Andrew Perlman, predicts AI will be "the most impactful technology the legal profession has ever seen." He thinks we'll eventually require lawyers to use AI—"at some point, we will start worrying about the competence of lawyers who don't use these tools.

The”

The question isn't whether AI will take over legal work. It's whether you'll be one of the lawyers who learned to work with it properly or one getting sanctioned for submitting fake case law to federal judges.

IN PARTNERSHIP WITH SELECTSOFTWARE REVIEWS

Get Matched With the Best HRIS/ATS Software, for Free!

Researching HR systems shouldn’t feel like a second job.

The old way meant hours of demos, irrelevant product suggestions, getting bombarded with cold emails and sales calls.

But there’s a better way.

With SelectSoftware Reviews, spend 15 minutes with an HR software expert and get 2–3 vendor recommendations tailored to your unique needs—no sales pitches, no demos.

SSR’s free HR software matching service helps you cut through the noise and focus only on solutions that truly fit your team’s needs. No guesswork. No fluff. Just insights from real HR experts.

Why HR teams trust SSR:

✅ 100% free service with no sales pressure
✅ 2–3 tailored recommendations from 1,000+ vetted options
✅ Rated 4.9/5 by HR teams and trusted by 15,000+ companies

Skip the old way—find your right HRIS/ATS in a new way, for free!

Get Your Free HR Software Matches

The Bottom Line

The future belongs to people who can harness AI while compensating for its blind spots. Those who get left behind are either avoiding AI entirely or trusting it like gospel.

Both approaches are career killers. Smart money is on the humans who learned to be better humans with AI, not despite it.