The one test behind every regime: net impression
Consumer law in most major markets does not ask whether a photo was edited. It asks whether the listing, taken as a whole, is likely to mislead an ordinary buyer about something that would affect their decision to buy.
In the US, the Federal Trade Commission's 1983 Policy Statement on Deception sets out three elements: a representation or omission likely to mislead, a consumer acting reasonably in the circumstances, and materiality, meaning the point would matter to the purchase. The FTC looks at the "net impression" of an ad, not each element in isolation. A small-print line under a photo does not cure a picture that says something different.
The EU's Unfair Commercial Practices Directive uses a similar structure. A practice is a misleading action (Article 6) if it contains false information, or presents true information in a way that deceives the average consumer, about the main characteristics of the product, and causes them to take a transactional decision they would not otherwise have taken. Article 7 covers misleading omissions: leaving out information the average consumer needs.
The UK, Australia and Canada all apply versions of the same idea. In practice this gives you a working test that travels well:
Would a reasonable buyer who opened the box say "this is what the photo showed me"? If the answer depends on what they were looking at (colour, size, condition, what is included) and the honest answer is no, the image is a problem, however it was produced.
Two consequences follow. First, an edit that changes nothing a buyer cares about is not misleading, even if it is heavy. Replacing a cluttered kitchen counter with a seamless white sweep changes every background pixel and misleads nobody. Second, a tiny edit can be misleading if it touches a material fact. Removing one scratch from a vintage watch face is a one-second fix and a straightforward misrepresentation of condition.
Edits ranked by legal risk
Most retouching sits comfortably on the safe side of the line. The edits that cause trouble cluster around a small number of material facts: colour, size, condition, what is included, and what the product does.
| Edit | What it tells the buyer | Risk |
|---|---|---|
| Removing dust, lint, fingerprints, sensor spots | Nothing they will not receive: the new item arrives clean | Low |
| Correcting white balance toward the item's real colour | A more accurate colour than the phone captured | Low |
| Replacing the background, adding a shadow | Nothing about the product itself | Low |
| Steaming out packing creases on a new garment | How the garment looks once unpacked and hung | Low |
| Pushing saturation or shifting hue for appeal | A colour the item does not have | High |
| Hiding scratches, stains or wear on a used item | A better condition than the one being sold | High |
| Showing props, accessories or set items that are not included | That they come with the purchase | High |
| Scaling a product up against a hand or a room | A size it does not have | High |
| Exaggerating a product's effect (skin, hair, cleaning, fit) | A result the product will not deliver | High |
| Re-drawing text, dials, serial numbers or markings | Details that are not on the real item | Depends on the category |
The pattern in the high-risk rows is that each one changes a fact a buyer would use to decide. Colour is the most common: it is a frequent cause of "not as described" complaints, and it is the edit most likely to be made casually, because a slightly richer red simply looks better in the grid.
The props row catches a lot of well-meaning sellers. A ceramic mug styled with a matching saucer, a lamp shown with its bulb lit, a backpack photographed full of gear: if the listing does not make clear what is and is not included, the photo is making a promise. A short "saucer not included" line next to the image is usually enough. A sentence buried in the description often is not, under a net-impression test.
How the main markets differ
The test is broadly shared, but the enforcement machinery and the specific rules around retouching differ. These are the regimes a cross-border seller is most likely to meet.
| Market | Core rule | Who enforces | Retouching-specific rules |
|---|---|---|---|
| United States | FTC Act Section 5 (unfair or deceptive acts), plus state consumer protection statutes | FTC, state attorneys general, private class actions in many states | None general; endorsement and influencer guidance applies to filtered content |
| European Union | Unfair Commercial Practices Directive, implemented nationally | National consumer authorities, coordinated through the CPC network | France requires a "photographie retouchée" label on commercial images where a model's body shape has been altered (since 2017) |
| United Kingdom | Digital Markets, Competition and Consumers Act 2024, Part 4 (in force from April 2025, replacing the 2008 regulations) | CMA (now with direct fining powers), Trading Standards; ASA for advertising | ASA guidance on beauty filters in ads; repeated rulings against over-retouched cosmetics ads |
| Australia | Australian Consumer Law, s18 (misleading or deceptive conduct) and s29 (false representations) | ACCC, state fair trading bodies | None general; covered by s18 |
| Canada | Competition Act deceptive marketing provisions | Competition Bureau | None general |
| Norway | Marketing Control Act | Consumer Authority | Since 2022, paid posts with a retouched body or face must carry a label |
Two shifts are worth knowing about. In the UK, the 2024 Act gave the Competition and Markets Authority power to decide consumer law breaches itself and fine businesses up to 10% of global turnover, rather than going to court first. In the EU, the 2019 Omnibus Directive required member states to make fines of at least 4% of annual turnover available for widespread infringements. Both moves make image accuracy a board-level risk for larger sellers rather than a customer service matter.
Advertising codes add another layer. The UK's Advertising Standards Authority has banned cosmetics ads where retouching overstated what a foundation or mascara does, and in 2021 ruled that social media filters which exaggerate a beauty product's effect are likely to mislead. Product photos on your own site are not always "ads" in the narrow sense, but the same reasoning is applied to them by consumer regulators.
AI-generated imagery has its own labelling questions, which we cover separately in AI image disclosure rules for e-commerce. The point here is narrower: labelling an image as AI-made does not make a misleading image acceptable. A disclosed fake colour is still a fake colour.
Where the risk actually lands: returns and chargebacks
For most small and mid-sized sellers, the realistic consequence of a misleading image is not a regulator's letter. It is a slow, expensive stream of returns, chargebacks and marketplace penalties, each of which is easier to win when the buyer can point to a photo.
In the EU and UK, distance sales already carry a 14-day right to cancel for most goods, so a buyer does not need to prove anything to send an item back. A misleading image turns that statutory right into a near-certainty. In the US there is no general federal right of return, but card networks give buyers a dispute route: Visa's "Not as Described or Defective Merchandise" reason code exists for exactly the case where the item received differs from what was shown. A seller fighting that dispute with a listing photo that shows a different colour is unlikely to win.
Marketplaces sit on top of both. Amazon, eBay and Etsy all treat "item not as described" as a seller-performance problem, and repeated claims affect account health and visibility. On a marketplace, the platform's own policy is usually stricter and faster than the law.
What a misleading image costs
- Returns on items that were never faulty
- Chargebacks that are hard to contest
- Marketplace account-health penalties
- Negative reviews quoting the photo
- In larger cases, regulator attention
What an accurate image protects
- A listing you can point to in a dispute
- Fewer "not as described" claims
- Reviews that say "exactly as pictured"
- A catalog that survives a marketplace audit
- A defensible position if a complaint escalates
The link between imagery and returns is covered in detail in how to reduce returns with better product photography. The legal framing adds one thing: when a return is caused by the photo, you are not just losing margin, you are building a record.
Used, vintage and one-of-a-kind items
Resale is where image accuracy and consumer law meet most directly. When a buyer is purchasing a specific item rather than a new unit from stock, the photo is not an illustration of the product. It is the product's condition report.
That changes what counts as a safe edit. On a new mug, removing a fingerprint is cleaning the photo; the buyer receives a clean mug. On a used mug, removing a chip is hiding a defect; the buyer receives a chipped mug. The same tool, the same three seconds, opposite legal positions.
A practical rule for resale catalogs:
- Fix the photo, never the object. Exposure, white balance, background, sensor dust and stray lint on the backdrop are photo problems. Scratches, stains, fading, pilling, chips and patina are the object.
- Show the flaws on purpose. A dedicated close-up of each significant defect is both persuasive and protective. Buyers who saw the flaw before purchase rarely claim it as a surprise.
- Do not let an AI tool "improve" condition. Generative re-drawing tends to produce a cleaner, newer-looking object. On a used item that is a misrepresentation, even if nobody intended it.
Retouchable's Studio handles this by offering two modes. Exact keeps the seller's own pixels and only corrects the light, so scratches and patina stay exactly as shot. Studio relight re-photographs the product and is labelled as AI in the interface; for used items it has a "keep wear" option, but because any re-drawn scratch is a likeness rather than the original, the app tells sellers to check condition or choose Exact. For a vintage piece, Exact is the conservative choice.
Fine markings, labels and text
One category of edit is easy to overlook because nobody does it on purpose: the accidental re-drawing of text and markings. Image models, upscalers and aggressive sharpening can all alter small lettering. A watch dial can come back with a garbled brand name. A coin's date can shift. A care label can sprout characters that do not exist. A serial number can lose a digit.
For most products this is a quality problem. For some it is a legal one. If buyers judge the item by its markings, such as watches, coins, trading cards, collectibles, electronics with model numbers, or anything sold on provenance, an image that shows different markings from the real item misrepresents a main characteristic. In the case of branded goods it can also raise trademark and counterfeit questions that are well outside the scope of a returns policy.
Before publishing any edited image of a product sold on its markings, zoom to 100% and compare every piece of text against the original photo. Thumbnails hide exactly the errors that matter.
Some tools build this in. Retouchable's Studio scans each photo before a relight and sends items judged by fine markings, such as a watch dial, a coin or a legible serial, to Exact mode so the original pixels are kept, unless the seller explicitly chooses otherwise. Whatever tool you use, the rule is the same: an item sold on its markings should be shown with its own markings.
If you source imagery from suppliers rather than shooting it yourself, the same accuracy questions apply, along with licensing ones. See the legal risks of using supplier product images.
A review process that keeps a catalog defensible
You do not need a legal department to keep product images on the right side of the line. You need a short, repeatable check that someone runs before an image goes live, and a record that it was run.
| Check | Question | Pass looks like |
|---|---|---|
| Colour | Does the image match the item under neutral daylight? | Side-by-side with the item, no visible shift in hue |
| Scale | Could a buyer guess the size correctly from the photos? | At least one image with a known reference, plus dimensions in the listing |
| Contents | Is everything visible in the photo included? | Props removed, or "not included" stated next to the image |
| Condition | For used items, are the flaws still visible? | Defects shown, ideally in their own close-up |
| Markings | Do text, dials and labels match the real item? | Checked at 100% against the original photo |
| Claims | Does the image show a result the product delivers? | No exaggerated before/after, fit or effect |
Three habits make this stick:
- Keep the original. Store the unedited photo alongside the published one. If a buyer or a regulator asks what was changed, you can show them. This is also your best evidence in a chargeback.
- Write down your edit policy. One page saying which edits are always allowed (cleanup, background, light toward true colour), which need a second look (anything touching colour or shape), and which are never allowed (hiding defects, changing markings). It gives freelancers and new staff a standard and gives you a document to point to.
- Audit by complaint, not by calendar. When a return or review mentions colour, size or condition, pull the listing images first. Complaint data tells you which products to fix far faster than a periodic review of the whole catalog.
None of this slows a catalog down much. Most edits pass every check without thought, because most edits are cleanup. The checklist exists for the small number that touch a material fact, which are also the ones most likely to turn into a return, a dispute or a complaint.