← All beauty signals
Trust signal

The Invisible Filter

Norway and France require labels on ads that retouch a body. The clearest UK ruling caught a dramatic skin-clearing effect, not a filter too fine to see.

What is it?

Two governments already require a label when an ad alters a body or a face, assuming a viewer can’t tell. The one real enforcement case found caught filtering the regulator itself called ‘dramatic’ — a detectable case, not proof a truly invisible filter has ever been caught.

Why now?

Norway and France passed disclosure laws years before generative tools existed, betting viewers cannot spot retouching unaided. Platforms now draw their own line only at ‘excessive’ filtering, tolerating a lighter layer beneath it. As of 2026, the EU adds a labelling duty, though not every AI touch triggers it.

Who’s building it?

Advertising regulators in Norway and France, ad platforms, and now the EU’s synthetic-content labelling regime.

Disclosure lawmakersAd regulatorsPlatform policy teamsFilter makersEU regime
The full read
Why now, who is building it, where the market goes, and what stays unsettled.
Get weekly →
01 · The market
Who’s in it
Lawmakers — in Norway and France, requiring a disclosure label.
Platforms — drawing their own line at ‘excessive’ filtering, not at detectability itself.
Shoppers — unable to separate a smoothed photo from an untouched one.
What’s at stake
Whether a viewer can ever reliably tell a subtly filtered face from an untouched one, disclosed or not.
Disclosure law assumes what enforcement has never actually tested.
Platforms quietly tolerate whatever sits beneath the word ‘excessive.’
What’s unsettled
Is ‘excessive’ a real threshold, or just a convenient one?
Does any regulator or platform test for undetectability at all?
Whether the EU’s hedge lets a subtle filter pass where a synthetic face can’t.
02 · How the demand matureda progression, not a timeline
Assumed blindness
the working premise behind two disclosure laws written before generative AI.
Detectable, not hidden
the one real ruling caught filtering the regulator called dramatic.
A threshold, not a ban
platforms restrict only ‘excessive’ filtering, tolerating an undisclosed lighter layer beneath it.
A statute with a hedge
the EU concedes not every AI touch needs disclosing.
03 · Where the market goesthe forks, not a prediction
→Either regulators eventually test a genuinely undetectable filter directly, or enforcement keeps catching only the filters that already look dramatic.
→Platforms could tighten ‘excessive’ into a real, testable definition, or leave it as a convenient line.
→The EU’s hedge could narrow toward disclosure, or stay wide enough to exempt most filters.
→A brand could simply keep its filtering beneath whatever line still counts as light.
Underneath every fork sits the same unmeasured question: not whether filtering happens, but whether any viewer, told or not, could ever have spotted it.
04 · The Filter-Trust deskadjacent signals, same gap
The Rendered Result
A generated ‘after’ image, still unruled.
The ASA upheld against filtered skincare results back in 2022.
Advertising Standards Authority
The Previewed Procedure
A rendered face already previews a surgical outcome.
A Wits University surgeon calls previews a surface-level estimate.
IOL / Sunday Tribune
The Unretouched Promise
Two brands now pledge to never generate bodies.
Aerie says any replicated body or person is unacceptable.
Marketing Brew
05 · The Wire
01
UK regulator upheld a complaint over a filtered skincare ad
Advertising Standards Authority ↗
02
Norway requires a label on any altered body or skin.
It's Nice That ↗
03
France requires a retouched-photo notice on slimmed or thickened models
NPR ↗
04
Douyin banned revealing outfits and excessive beauty filters
Sixth Tone ↗
05
UK ad regulator: advertisers remain responsible for AI tool outputs
Advertising Standards Authority ↗
06
EU requires clear AI-content labels in key cases from 2 August 2026
European Commission ↗
The panel on this signal
They agreed on
–Disclosure law already assumes viewers can’t detect filtering — enforcement so far has only caught the obvious cases.
–Norway and France both legislated years before generative AI.
–Platforms only restrict filtering once it crosses into what they call ‘excessive.’
The visual-perception researcher disagreed on
–Untrained viewers routinely miss far more obvious retouching than ‘excessive’ implies, so the real gap sits well below that line.
Know more about the method →
✦How this signal was built
01 · Panel
A Delphi panel of 4 AI analysts, each reasoning independently.
→
02 · Debate
They challenge and revise each other’s reads over two rounds.
→
03 · Evidence
Every claim is tied to a cited, linked source on The Wire.
→
04 · Synthesis
A moderator writes it into one read, disagreement and all.