Does every ad need an AI label?
On 22 October, EASA released a guide: not every AI ad needs a label. Three questions decide. And a label will not save a misleading ad.

Does every AI ad really need an “AI” badge? The European Advertising Standards Alliance says no.
On 22 October 2026, EASA published a non-binding guide. The core point in one sentence: use a label only when the disclosure actually helps the customer — not for every small AI touch.
The three questions
EASA looks at three things:
- How much of a role did AI play?
- Does the result look real?
- Is it relevant to the purchase decision?
The higher all three, the more likely a label is needed. Little AI, clearly fictional, or irrelevant to the customer? Then often none.
What a label does not fix
Misleading stays misleading. Even with an “AI” sticker. A label is not a substitute for honest advertising.
What you should do in practice
- Sort your AI creatives against the three questions.
- Where a label is needed: make it clear, immediately visible, and close to the ad.
- Factor in Meta’s automatic variants: they often appear after you go live.
What we review at ADGENTIC
Nobody checks dozens of AI variants by hand. We review ads before they go live, and the variants Meta then builds itself.
Fewer labels. More sense. And ads that still hold up without a badge.
Further reading
- EU AI Act for SMEs: what you actually need to do from August 2026
- Can I clone models with AI and use them in ads?
- Meta’s AI is tinkering with your ads. You are liable.
How ADGENTIC's AI agents plan, run and optimise campaigns is explained in the overview of the ADGENTIC agents.
Sources: EASA, 22 October 2026 (easa-alliance.org), Lewis Silkin (lewissilkin.com)

