Use case · AI content creation

AI photo retouching at scale, within the rules.

AI retouching can clean, colour-correct and finish thousands of images to one standard, but scale repeats a dishonest edit as easily as a good one. The working rule: clean what is temporary, keep what the customer will get. Changing body shape triggers labelling laws in France and Norway, and exaggerating a cosmetic result is misleading advertising almost everywhere.

France: the 'photographie retouchée' label

Since 1 October 2017, commercial photographs in France showing models whose body silhouette has been made thinner or thicker with image-editing software must carry the words 'photographie retouchée'. The rule sits in the Public Health Code, implemented by Décret n° 2017-738, and the fine is €37,500, which can rise to 30% of the advertising spend. It covers advertising images of models in print, on posters and online. The law speaks of image-editing software without distinguishing tools, so a silhouette reshaped with a generative tool is likely to count too. Skin clean-up alone does not trigger the label; changing the silhouette does. Build the check into the workflow for every image that will run in France.

Norway: the retouched-body mark

Norway went further. Since 1 July 2022, its Marketing Control Act has required advertising in which a person's body shape, size or skin has been changed by retouching or other manipulation to carry a standard mark set out in government regulations and enforced by the Norwegian Consumer Authority. It applies to paid posts by influencers as well as brand campaigns, and to video as well as stills. Filters and AI edits count when they change a real person's body, though the Consumer Authority's view is that a body generated from scratch is not covered. For a global campaign, the practical answer is a labelled or unaltered version for Norway, and a record against every image of whether bodies or skin were altered, so the right version reaches the right market.

Colour accuracy for products

For product imagery, the most important retouch is often invisible: getting the colour right. A jumper that arrives a different shade from its photograph is a return and a complaint. Capture or render against a colour reference, calibrate the screens that make the call, and compare with the physical sample under a standard light source. Agree a measurable tolerance, a Delta E limit, rather than approving by eye on a laptop. Then police the automation: batch settings that brighten, warm or add contrast for appeal can quietly move an entire range off its true colour, and nobody notices until the returns come back.

Never exaggerate a cosmetic effect

Cosmetics advertising has the strictest line. In the EU, claims for cosmetic products must meet the common criteria in Regulation (EU) No 655/2013, including honesty: presented performance must not go beyond the supporting evidence. In the UK, the ASA has repeatedly ruled against beauty ads whose production techniques or retouching exaggerated the result, and in 2021 it ruled that social media filters which exaggerate a product's effect are misleading. So retouch the skin around a mascara, never the lashes it claims to lengthen; never smooth the lines an anti-ageing cream claims to reduce; and use lash inserts only to replace missing lashes, never to add length the mascara cannot.

Updated 25 September 2026

Questions

AI photo retouching, answered.

Does removing a blemish count as retouching under these laws?

Not under France's rule, which targets changes to the body's silhouette. Norway's mark covers changes to body shape, size or skin, so skin work can fall inside it. Keep a written house standard of allowed edits per market, and record which images had skin or body work done, so the right version runs in each country.

Can AI retouching make product colours more accurate?

Yes, if it is aimed at a measured reference rather than at appeal. Automated correction against a colour target and the physical sample can bring a whole range into line quickly. The same tools set to enhance will push colour away from the product, so keep the target and the tolerance in the brief.

How is retouching kept consistent across a large range?

By writing the standard down and enforcing it where images are made. In Synthetic White, the image style sits in the brand hub and is enforced at generation, product-lock holds product colours and labels to the reference, and approvals carry roles, sign-off and an audit trail, so every retouched image has a record of who approved it.

Get started

See your AI studio generating this week.

Thirty minutes. A live studio in your colours, your brand hub loaded, a real campaign brief.

See plans