I was at the DA4S Alliance for Sciences conference in Anaheim a couple weeks ago, and I sat in on a legal panel about AI. I went in expecting a snooze. I came out doing research for days afterward, because what these lawyers laid out wasn’t theoretical, it was real businesses, real lawsuits, and real dollars lost by people who dipped their toe into AI without checking if the pool had a lifeguard.

Here’s the contradiction nobody talks about at dinner parties: AI is out here helping develop cancer drugs faster than any human team ever could, running 18,000 test permutations while we’re still arguing about ChatGPT prompts. And in the very same breath, it can quietly write your business a subpoena. Miracle and menace, holding hands.

I’m not here to scare you off AI. I use it daily and I’m not going back. But I am going to walk you through the 5 ways it’s actually landing small businesses in court right now, so you don’t find out the hard way, in front of a judge, with your lawyer billing you by the confused hour.

Risk #1: Your Chatbot Talks. You Get The Bill.

In 2024, a customer asked Air Canada’s chatbot about a bereavement refund. The bot, full of confidence and zero accuracy, said “go ahead, book your flight, apply for the refund after.” Total fiction. The customer believed it, flew, applied, got denied, and sued.

Air Canada’s defense? “It wasn’t us, it was the chatbot.” Adorable. The court did not agree, and neither would I if I were the judge.

**The lesson:** your chatbot, your problem. Doesn’t matter if it’s AI, a glitch, or an intern who had one too many espressos and started freelancing customer service. If it’s got your logo on it, you own every word that comes out of it.

Risk #2: “The AI Did It” Is Not A Defense. It’s A Confession.

An online tutoring company built a hiring agent and, somewhere in its instructions, it decided to screen out women over 55 and men over 60. Not just ageist. Unevenly sexist too, with different cutoffs for men and women, which is a genuinely impressive amount of discrimination to pack into one algorithm.

The Equal Employment Opportunity Commission came knocking. The company lost, obviously.

**The lesson:** if it’s your agent, your algorithm, or your screener, it’s your fault. Blaming the AI is like blaming your car for speeding. You were driving.

Risk #3: It Lies With Total Confidence. That’s Defamation Waiting To Happen.

A journalist asked ChatGPT to summarize a real firearms lawsuit for publication. The AI, instead of summarizing, invented an entirely fictitious claim that a radio host had committed embezzlement and fraud. The journalist published it. The radio host sued.

The case got dismissed, but not because AI was innocent. It got dismissed on a technicality nobody could untangle: who exactly was at fault, and how do you put a dollar figure on a lie a robot told? Messy, expensive, and nobody actually won.

I deal with this hallucination habit constantly, and not just in business. I feed my mom’s years of medical tests into AI to help me prep talking points for her doctor. Every single time, I tell the doctor upfront, “this is AI generated, please fact check it,” because every single time, something in there is dialed up to a level of alarm that not one actual doctor has ever agreed with. AI is confident. AI is not always right. Those are two very different superpowers, and only one of them should be running your business.

**The lesson:** never publish, present, or act on AI output without a real human giving it the skeptical once-over first.

Risk #4: “I Found It Online” Stopped Being A Defense A Long Time Ago

Anthropic, the company behind Claude, got hit with a $1.5 billion lawsuit for using 500,000 printed books as training content without asking anybody’s permission first. They lost. And that’s just the headline case. Over 130 lawsuits are currently pending against AI companies, filed by names you’d recognize: The Atlantic, Forbes, The Guardian.

**The lesson:** if you’re using AI to write your blogs, generate your images, or build your marketing content, and that AI was trained on someone else’s copyrighted work, the liability doesn’t land on the AI company. It lands on you, the business that hit “publish.”

Risk #5: What You Type In Doesn’t Always Stay In

Back in 2023, Samsung engineers pasted proprietary source code and confidential meeting notes into ChatGPT, just trying to get some quick help debugging. They had no idea that data could become part of the AI’s permanent training set. Samsung’s response was swift and dramatic: a company-wide ban on generative AI tools, full stop, every department.

**The lesson:** vendor contracts almost never cover what happens to your data once it enters an AI tool. Assume anything you type in could end up somewhere you never intended it to go, and act accordingly.

Your SHIELD Formula

I’m not telling you to run screaming from AI, unplug the router, and go back to fax machines. I use it every single day and I’m not giving it up. But I am telling you to stop using it naked, with zero protection. So here’s your SHIELD: (I learned how to do these acronyms for my TV appearances lol!)

**S**creen every AI tool and agent you use for liability language before you touch it.

**H**uman eyes on everything AI produces, before it goes public, gets sent, or gets acted on.

**I**nsurance. Call your insurer this week and ask specifically about AI-related liability coverage.

**E**verything documented. Your prompts, your edits, your dates. Several businesses in these lawsuits only survived because they had a paper trail to prove their intent.

**L**iability reviewed in every vendor contract, before you sign, not after you get served.

**D**ata assumed exposed. Whatever you type into an AI tool, assume it could end up somewhere you never meant it to.

AI is a dangerous unknown animal that’s also, in the same breath, saving lives and changing the world. The businesses that win this round won’t be the ones avoiding AI. They’ll be the ones using it two steps smarter than everybody else in the room.

Don’t let a chatbot’s confident nonsense become your legal bill.

Want to reposition your messaging to grow your leads? Follow me on X, friend me on Facebook,watch my Podcast onYouTube or connect with me on LinkedIn –and let’s talk.