Overstating what your AI does is the fastest way to get sued. Say only what's true.

The pressure is easy to feel. Everyone in your market is putting AI on everything, the deck looks stronger with it, and the marketing team wants to say the product is AI-powered end to end. It is tempting to round up, to describe what the AI will do soon as though it does it now, to let a partner's technology sound like your own. Almost everyone is nudging the story in that direction. And that is exactly why the regulators started paying attention.

So here is the shift worth naming. Overselling your AI has moved from harmless enthusiasm to real exposure, the legal and reputational kind. This is not a reason to hide what your AI does. It is a reason to describe it precisely, because precision is now both the safe position and, increasingly, the credible one.

Is overstating your AI actually a legal risk?

It is, and there are now cases to prove it. In 2024 the US Securities and Exchange Commission brought its first "AI-washing" actions against two investment firms for making false or misleading claims about their use of AI. They paid a combined US$400,000 in civil penalties. Since then the enforcement has widened to more companies, including firms charged with overstating what their AI products could do, and other agencies have opened parallel work. The pattern is set: if you claim it, you may be asked to prove it.

One honest caveat, because it matters for how much weight to give this. The clearest enforcement so far has been in the United States, and it has hit public companies and regulated financial firms hardest. But the principle behind it, that you must not mislead the market about your product, exists in most places, including Australia through consumer and corporate law. So treat this as a direction of travel, not a distant foreign quirk. It is a close cousin of the discipline behind being honest about what AI is really delivering rather than what the hype promises.

Why is the temptation to oversell so strong, and so risky?

Because the short-term reward is real and the cost is delayed. Saying more sounds better in the pitch, wins attention, and keeps pace with rivals who are also rounding up. So the whole market drifts toward the same inflated language, and it feels normal because everyone is doing it. That is precisely the condition regulators look for, and precisely when an overstatement becomes a claim you cannot back. The gap between what you said and what your AI does is the whole risk.

The lower-altitude question is how to make the AI story sound as impressive as possible. That is the instinct that gets firms into trouble. The better question is how to make it both true and compelling, because those are not opposites. A precise, confident account of what your AI really does, and where a human or a partner sits in the loop, reads as more credible than a vague "AI-powered everything," not less. It is the same move as holding your AI claims to the standard a board would want to defend.

So how do you tell the AI story safely?

You make honesty the policy, and you make precise language the habit across everyone who describes your product. The aim is a story that is both true and strong. Work it in this order:

1. Say what it does today, not what it might do. Describe the current capability plainly. Keep the roadmap clearly labelled as future, never as present fact.
2. Name the human and the partner. Be clear where people or third-party technology sit in the loop. Claiming their work as your autonomous AI is a common and avoidable trap.
3. Match the claim to the evidence. Before a claim ships, check it against what the technology can actually demonstrate. If you cannot show it, do not say it.
4. Align marketing, sales and investor language. The exposure spans all of them. Make sure they tell the same, accurate story.
5. Have one owner for AI claims. Give someone responsibility for checking that what the business says about its AI is true. Ambiguity is where overstatement grows.

What does this make possible?

An AI story that holds up, and quietly wins because of it. When you say only what is true, you never have to walk a claim back, and customers learn that your word is reliable. As overclaiming gets punished and the market grows tired of "AI-powered everything," a precise, honest account of what your technology really does stands out as the credible one. Told straight, your restraint becomes an advantage while others are exposed. That is a stronger position than a headline claim you cannot actually stand behind.

Picture the market a year on, as the inflated language starts to cost the firms using it. Your AI story needs no correction, because it was accurate from the start. Customers trust it, regulators have no quarrel with it, and it reads as more confident than the hype around it, not less. You resisted the easy exaggeration and told the truth well, and in a market learning to distrust AI claims, the business people believe is the one that wins.

About the author

Thomas Green

British technology futurist, AI keynote speaker and advisor. Thirty years across enterprise technology and AI strategy, helping leaders navigate the future of work. The futurist who died.