Your Users Don't Want AI — Now You Have to Label It

What if the hardest design problem AI creates isn't about generation or automation — but about honesty?

The EU has just published new guidelines on how AI-generated and AI-powered content should be labelled. And the headline for designers is blunt: that little sparkle icon you've been dropping next to AI features? It's not enough.

What Happened

The new EU guidelines spell out when and how AI must be disclosed in digital products. This covers both content that's been generated by AI and features that are powered by it under the hood. The rules push for standardised, clear labelling — not the vague iconography most product teams have been getting away with.

This isn't a suggestion. For anyone shipping products into the European market, it's a regulatory expectation that's going to shape interface design in concrete ways.

At the same time, something else is becoming painfully clear: most people don't actually want more AI in their products. There's a real and widening gap between what tech leadership is enthusiastically building and what users are asking for. The assumption that everyone's desperate for AI features is looking increasingly shaky.

The Bigger Picture

Here's the awkward situation designers now find themselves in.

On one side, companies are racing to inject AI into everything — search, email, photo editing, messaging, shopping, you name it. The internal pressure to ship AI features is enormous. On the other side, users are largely indifferent or actively sceptical. And now, regulation is saying you can't just sneak AI in quietly — you have to tell people it's there.

This creates a genuinely interesting design problem. You're being asked to prominently label something that your users probably didn't ask for in the first place.

Think about what that means in practice. Every AI disclosure is a small moment of friction. A label that says "this was generated by AI" isn't neutral — it carries baggage. For some users it signals unreliability. For others, laziness. For many, it's just noise they didn't need.

The temptation will be to treat AI labelling as a compliance checkbox — bury it in settings, make the text tiny, use the most anodyne language possible. But that's exactly the wrong instinct. If the goal is genuine transparency, designers need to find ways to disclose AI that actually build trust rather than erode it.

This is where our craft matters. The difference between a label that feels informative and one that feels like a warning is pure design. It's tone of voice, placement, timing, visual weight. It's everything we're supposedly good at.

What to Actually Design For

The EU guidelines are pushing designers toward a few specific shifts worth paying attention to:

Distinguish between types of AI involvement. There's a difference between content that was fully generated by AI, content that was AI-assisted, and features that use AI in the background. Your labelling should reflect that distinction — not collapse everything into a single sparkle.

Context matters more than icons. A generic icon doesn't tell users what they need to know. Was this image generated? Was this summary written by a model? Is this recommendation personalised by an algorithm? Each of these deserves its own clear, contextual disclosure.

Make it useful, not decorative. The best AI labels won't just say "AI was here" — they'll tell users what that means for them. Can they trust this output? Can they edit it? Was their data involved? These are UX questions, not legal ones.

If you're working on any product with AI features shipping into Europe, now is the time to audit your current approach to AI disclosure. Map every touchpoint where AI is involved and ask yourself: would a user understand what's happening here without our label? And with it?

The Uncomfortable Truth

I think the deeper lesson here isn't really about labelling at all. It's about the gap between what we're building and what people need.

When users consistently say they don't want more AI, and regulators step in to demand transparency about AI, that's two signals pointing in the same direction. It's not that AI is useless — far from it. It's that the current approach of cramming AI into every feature and hoping users will be impressed is running out of road.

We're the ones who should be asking "does this feature actually serve the user?" before we ask "how do we label it?"

Takeaway

Audit your product's AI touchpoints this week. For each one, ask two questions: does the user know AI is involved here, and — honestly — did the user want AI involved here? If the answer to the first is no, you've got a compliance problem coming. If the answer to the second is no, you've got a design problem already. The labelling guidelines are a forcing function, but the real opportunity is using them as an excuse to have the harder conversation about whether your AI features are earning their place.