June 17, 2026
There’s a move I see founders make when they deploy AI on the front line of their business, and it costs them more than they realize. They coach the bot to sound human. Give it a name. A fake headshot. A breezy “hey, it’s Sarah from the team!” They think they’re removing friction. What they’re actually doing is planting a landmine under their own trust.
Let me say the quiet part out loud. Your AI shouldn’t pretend to be a person. And the reason isn’t soft. It’s strategic, and it’s about to be legal.
The instinct is understandable. You want the experience to feel warm, not robotic. So you disguise the machine to make the conversation smoother. The logic feels right: humans trust humans, so make the bot pass as one.
Here’s the flaw. You’re optimizing for the first ninety seconds of the interaction and ignoring the entire relationship that comes after. Smoothness up front, suspicion forever.
Customers find out. They always find out. The reply comes back in four seconds at 2am. “Sarah” answers thirty tickets in a minute. The cadence is a hair too perfect. People are pattern-matching machines, and they clock it.
And the moment they realize they were talking to a bot dressed up as a human, something specific happens: they don’t just feel like they hit a robot. They feel deceived. That’s a different, heavier emotion than “oh, this is automated.” Deception is personal. It tells the customer you were willing to trick them for your own convenience, which makes them wonder what else you’d shade. You traded a little short-term smoothness for a long-term tax on every interaction that follows.
Trust is the most expensive thing to rebuild in any business. You do not want to spend it on a disguise you didn’t even need.
This used to be a brand-judgment call. As of 2026, the ground is moving underneath it.
The EU AI Act’s transparency rules take effect August 2, 2026, and they require that people be told when they’re interacting with a machine, full stop, if you serve EU customers. California’s companion-chatbot law went live January 1. The FTC already treats hiding the bot as potentially deceptive. And per legislative trackers, more than two dozen US states have chatbot bills in motion.
The specifics vary and some carve out plain customer-service bots, so this isn’t legal advice. But read the direction, not the fine print. The regulatory arrow points one way: disclose. Building your customer experience on a disguise that the law is actively dismantling is not a foundation. It’s a liability with a countdown on it.
Here’s the part founders miss. Disclosure isn’t the cost of doing this right. It’s the advantage.
When your AI opens with “Hi, I’m the AI assistant for [brand], here to get you sorted fast,” you’ve just done three things at once. You set an honest expectation. You signal you have nothing to hide. And you let the customer relax, because nobody’s been tricked, so nobody’s on guard. Transparency isn’t the weakness in the system. It’s the whole game.
If you’ve got AI touching customers, run this today:
Disclose at the top, every time. First message of every conversation names it as an AI assistant. Not buried in a tooltip. Right at hello.
Kill the fake human costume. No invented name, no stock-photo face, no pretending a person is typing. Give it a clear assistant identity instead.
Build the human exit ramp. Make “talk to a real person” obvious and fast. AI that can’t hand off cleanly is where trust actually breaks, not the disclosure.
Be honest with the buyer too. When you sell or pitch this, frame what the AI does and where a human steps in. Honest framing for the buyer, full disclosure for the customer. Same principle, both ends of the deal.
Audit the edges. Voice agents, SMS flows, DM auto-replies. The disguise usually hides in the channels you set up once and forgot.
This is the IVM through-line in everything we build. Automation without architecture creates chaos. Architecture creates freedom. An AI system designed to deceive is a chaos machine wearing a friendly mask. One designed to be transparent compounds trust every single time it runs.
You don’t need your AI to pass as human. You need it to be useful, fast, and honest about what it is. That combination is what actually converts.
Want a quick 10-minute look at where your AI touches customers and whether it’s set up to build trust or quietly burn it? Hit reply.
— Gabi
Intentional Visionary Media designs AI growth systems for founders who’d rather build leverage than collect tools.
No posts
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.