The EU now requires AI to say it is AI. It cost me nothing.

Since August 2, AI in Europe has to tell you it is AI. If a customer is chatting with an AI, the AI has to say so. If content was made by AI, it has to be marked as AI-made. Those two rules are now law for anyone using AI in their business, and a lot of people spent the week nervous about them. I run a whole organization of AI agents, so I went through my setup to see what the new rules would cost me. The answer was nothing. Here is why.

What actually changed on August 2?

The EU’s big AI law has been arriving in pieces for a while, and this month two pieces landed at once. The first piece is the openness rules, and those are live now. In plain words: no pretending an AI is a human. A chatbot talking to your customers has to disclose that it is an AI. Content that an AI made has to be marked as AI-made.

The second piece is a postponement. The heavier requirements, the ones for AI making serious decisions about people, things like hiring and other calls that change someone’s life, were pushed back to December 2027. So the message from the EU is pretty simple. Be honest now. The hard requirements come later.

What did the new rules cost my agent organization?

Nothing. I had nothing to change, and not because I am especially clever about regulation. It is because hiding the AI was never part of the plan. My whole thing, the reason you are reading this, is showing that AI agents do the work. I post about what my agents built, what they broke, and what I had to fix. The openness the law now demands is the same openness I have been publishing all along.

Think of it like a restaurant with an open kitchen. When the inspector shows up, nothing changes, because everyone could already see the stove. That is what this week felt like. The inspector arrived, looked around, and everything was already visible.

Why was the honest version always the better build?

Here is the part I actually believe, separate from any law. An AI that pretends to be human is a trick, and tricks have a shelf life. The first time a customer finds out they were fooled, you lose more trust than the trick ever earned. An AI that says “I am an AI, here is what I can do for you” sets an honest expectation and then either meets it or hands over to a human.

So when people ask what it takes to comply with rules like these, my answer is: build it honest from day one and compliance is a side effect. The businesses scrambling this month are mostly the ones that built the trick.

What should a small business owner do now?

If you use any AI that talks to customers or makes content, two checks. Does the AI say it is an AI? Is the AI-made content marked as AI-made? If both answers are yes, this law is a non-event for you, like it was for me. If either answer is no, fix it now, not because of the fines, but because the honest version was always the better product. Now it is also the legal one.

I build AI agents that run all parts of my online business. Check them out at pelleforsman.com.