By Chuck Gallagher — Business Ethics Keynote Speaker and Trainer
TL;DR: Chuck Gallagher, AI ethics speaker and author, examines Arkansas’s consumer-protection lawsuit against Snap — and what happens when a product built for compulsion attracts the people who exploit it.
In 2011, three Stanford students built an app around a single promise: send a photo, and it vanishes. No record. No permanence. Nothing your parents, your teachers, or some future employer could ever dig up. The idea, “What your parents don’t know won’t hurt ya.” They called it Picaboo first, then Snapchat. Several venture capitalists who heard the pitch shrugged; a few assumed it was a sexting app and passed. Teenagers understood it instantly. By 2014, the median Snapchat user was eighteen years old. Today the platform reports more than 400 million daily users.
That was the appeal from day one. Disappearance. Freedom from the receipt.
Now look at where that promise landed. On June 23, 2026, Arkansas Attorney General Tim Griffin filed suit against Snap, Inc. in Phillips County Circuit Court (Cause No. 54CV-26-251). The State alleges Snap marketed Snapchat to parents as safe and family-friendly while shipping features it knew exposed minors to harm: disappearing messages, the Snap Map location feature, age checks built on self-reported birthdays, and what the complaint calls an untested My AI chatbot that delivers dangerous and inappropriate advice. These are allegations. Snap has not been found liable. But the pattern they describe is one I have watched repeat for years.
What Was Snapchat Actually Designed to Do?
Here is the part people miss. The same feature that protected a shy teenager from permanent embarrassment also protected the adult who screenshotted her photo before it disappeared. As one of the attorneys on the case framed it, disappearing messages don’t disappear for the predator who already saved them. They disappear for the parent, the prosecutor, and the jury. The architecture cuts both ways. It always did.
The State’s deeper claim is about compulsion. Arkansas alleges Snap built dopamine feedback loops modeled on slot-machine mechanics — streaks, vanishing content, the constant pull to open the app and chase the next hit. As an AI ethics speaker and author, I’ll tell you the uncomfortable truth plainly: features that exploit a teenager’s craving for approval and fear of exclusion are not accidents. The complaint alleges they were calculated. Designed in. Built so a developing brain keeps coming back.
Why Does the My AI Chatbot Matter So Much?
Because it is the newest layer on an old foundation. Snap dropped a generative chatbot into a product already optimized to hold a minor’s attention. Arkansas calls it untested and alleges it can hand a child dangerous or inappropriate guidance. Whether that proves out in court is one question. The design question is simpler. If you bolt an unpredictable AI onto a platform engineered for compulsive use by kids, you have stacked one risk on top of another and called it a feature.
And here is where the rationalization usually lives. Growth first, safety later. Our controls are good enough. Parents are responsible. The edge cases are rare. I have heard every version of that sentence. I once told myself a version of it too, in a different context, years ago — and I learned what it costs. The edge cases are not rare. They are the people who were always going to show up the moment you build something compulsive and point it at the young.
What Is Arkansas Actually Asking For?
The lawsuit brings claims under the Arkansas Deceptive Trade Practices Act for deceptive and unconscionable practices, along with public nuisance and unjust enrichment. The State seeks injunctive relief — meaningful age verification, parental controls a child cannot override, and structural changes to the most dangerous features — plus civil penalties of $10,000 per knowing violation, damages, restitution, and abatement. It is one of a growing number of enforcement actions aimed at how platforms treat children; the same firm representing Arkansas recently won a $6 million addiction verdict against two other social-media companies. I won’t predict the verdict. Litigation is its own animal. But the lesson doesn’t wait for a judge.
What Should Any Company Building for Minors Take From This?
As an AI ethics speaker and author, the guidance I give boardrooms is plain: treat anything a child can touch like a regulated product. Age assurance with real friction, not a self-reported birthday any twelve-year-old can fake. Safety-by-default settings, so protection isn’t something a parent has to discover and switch on. Documented red-team testing for any chatbot before it ships, not after a complaint arrives. Incident reporting that actually works. None of this is exotic. It is the difference between meaning safe and engineering safe.
There is a useful baseline already written down. The European Union’s AI Act, in Article 50, requires that people be told when they are interacting with an AI rather than a human, and that rule takes effect August 2, 2026. You don’t have to operate in Europe to adopt it. Telling a kid plainly that the friendly voice in the app is a machine is not a compliance burden. It is basic honesty.
That’s the whole thing, really. When you design for compulsion, you eventually attract the people who exploit compulsion. Then you spend years acting surprised. The choice to build it that way was made long before the lawsuit. So was the consequence.
Frequently Asked Questions
Is Snapchat being sued over its My AI chatbot?
In part, yes. On June 23, 2026, the Arkansas Attorney General sued Snap, Inc. in Phillips County Circuit Court (Cause No. 54CV-26-251), and the complaint specifically alleges that an “untested” My AI chatbot can deliver dangerous and inappropriate advice to minors. That claim sits alongside broader allegations about disappearing messages, location features, and weak age verification. These are allegations, and Snap has not been found liable.
What laws does the Arkansas lawsuit against Snap rely on?
The State brings claims under the Arkansas Deceptive Trade Practices Act for deceptive and unconscionable trade practices, along with public nuisance and unjust enrichment. It seeks injunctive relief, civil penalties of $10,000 per knowing violation, damages, restitution, and abatement.
Why was Snapchat originally popular with teenagers?
Snapchat launched in 2011 on the promise of disappearing photos — messages that left no permanent record. Teenagers embraced it as an escape from the curated, permanent feeds of Facebook and Instagram and from adult surveillance. That same disappearing-message design is now central to the allegations against the company.
What does the EU AI Act require for chatbots?
Article 50 of the EU AI Act requires that AI systems built to interact with people be designed so users are told they are dealing with an AI rather than a human, an obligation that takes effect on August 2, 2026. As an AI ethics speaker and author, I treat that disclosure standard as a sensible baseline for any company deploying a chatbot, inside Europe or not.
How can a company reduce the risk of harming minors with its product?
Treat any product minors can access like a regulated one: age assurance with real friction, safety-by-default settings, documented red-team testing for AI features before launch, and working incident reporting. The aim is to build safety into the product rather than market it as safe after the fact.
For more than two decades I’ve helped boardrooms and leadership teams see the line between meaning well and building well — the place where a design choice quietly turns into a consequence. If your organization is putting AI in front of customers, employees, or kids, that line is exactly where the risk lives. Chuck Gallagher brings this conversation to leadership sessions and main-stage events across the country. To talk about your next one, visit ChuckGallagher.com.
Five Questions for Reflection
1. Where in your own product or service does “growth first, safety later” quietly govern the roadmap?
2. If a single feature can protect a vulnerable user and a predator with the same design, who did you actually build it for?
3. What would honest age assurance look like in your business — and what revenue are you afraid it would cost?
4. Before your company deploys an AI feature, who is responsible for red-teaming it, and what happens if they say no?
5. When the consequences of a design choice finally surface, will your team be able to say it was surprised — or only that it acted surprised?
