# AI-Generated Product Photos: Risks, Rules and Disclosure

URL: https://ecomchannels.com/insights/ai-generated-product-images-rules-risks/
Category: AI graphics & visual commerce
Updated: 2026-10-30

> AI can create product scenes in seconds, but Australian Consumer Law and marketplace rules still apply. The risks, the disclosure rules and a safe workflow.

AI-generated product images are allowed on most major marketplaces, but they carry real risks: Australian Consumer Law applies to images exactly as it does to words, and Amazon and Google now have specific metadata rules for AI-generated content. The safe approach is to use AI to change the setting around a real product photo (background, scene, lighting, cropping), never to change the product itself, and to keep a person approving every image before it goes live.

## What AI image tools now do

Generative AI has moved quickly from novelty to standard seller tooling. Google's free Product Studio can [create images, change and remove backgrounds, increase image resolution and generate product videos](https://support.google.com/merchants/answer/13708167?hl=en) inside Merchant Center. Amazon Ads offers AI image generation to advertisers, and eBay's app lets sellers swap a photo background for a solid colour or an [AI-generated scene](https://www.ebay.com.au/sellercentre/create-a-great-listing/photo-tips).

These tools cut the cost of lifestyle and seasonal imagery sharply. They also make it easy to publish an image that doesn't match the product in the box.

## Risk 1: misleading consumers

The Australian Consumer Law prohibits misleading or deceptive conduct, and the ACCC is explicit that this covers [images and descriptions of what is offered](https://accc.gov.au/consumers/advertising-and-promotions/false-or-misleading-claims), on websites and any other platform. Two points from the ACCC's guidance matter most for AI imagery:

- **Intent doesn't matter.** It [makes no difference whether a business intends to mislead or not](https://accc.gov.au/consumers/advertising-and-promotions/false-or-misleading-claims). An AI tool that "improves" a fabric texture, adds a feature, or changes the scale of a product relative to a room can create a false impression even if nobody meant it to.
- **Images in context count.** The ACCC's own case study is about images: Samsung was ordered in 2022 to pay [$14 million in penalties](https://accc.gov.au/consumers/advertising-and-promotions/false-or-misleading-claims) over ads showing phones used in pools and the sea, when pool and sea water could damage them.

Common AI-image failure points include altered colours or finishes, extra or missing components, wrong proportions, generated props that look included, and in-use scenes that imply a capability the product doesn't have (outdoor use, waterproofing, load capacity).

Misleading images also cost money directly. In Salsify's 2024 survey, [45% of shoppers had returned an item in the past year because of incorrect product content](https://www.salsify.com/hubfs/2024/Content/Salsify%20Consumer%20Research%20Report%202024.pdf), including images that didn't match the product.

## Risk 2: breaking marketplace rules

Marketplace rules apply regardless of how an image was made, and some now address AI directly:

- **Amazon** requires all images to accurately represent the product and match the title. If an image includes [photorealistic AI-generated people](https://sellercentral.amazon.com/help/hub/reference/external/G1881?locale=en-US), sellers must add the IPTC keyword `contains-synthetic-performer` to the image metadata before upload so Amazon can add a disclosure where applicable.
- **Google Merchant Center** requires that [all images created using generative AI contain metadata indicating they were AI-generated](https://support.google.com/merchants/answer/6324350), such as the IPTC DigitalSourceType "TrainedAlgorithmicMedia" tag, and tells merchants not to strip that metadata.
- **eBay** suggests using AI-generated backgrounds for [only a few photos](https://www.ebay.com.au/sellercentre/create-a-great-listing/photo-tips), with backgrounds that don't distract from the item. Its picture policy separately bans photos that don't accurately represent the item.

A practical consequence: image pipelines that compress or re-save files can strip metadata. If Google or Amazon require an AI tag, check it survives your export process.

## Risk 3: trust

Shoppers notice, and care. Getty Images' 2024 research, drawing on more than 30,000 adults in 25 countries, found [almost 90% of consumers globally want to know whether an image has been created using AI](https://newsroom.gettyimages.com/en/getty-images/nearly-90-of-consumers-want-transparency-on-ai-images-finds-getty-images-report), and that people feel less favourably towards brands using AI-generated visuals to create people or products. The distinction is useful: an AI-generated kitchen behind a real photo of your kettle is a different proposition from an AI-rendered kettle.

## Australian AI guidance points the same way

The federal government's Voluntary AI Safety Standard, updated in October 2025 as the Guidance for AI Adoption, isn't law, but its guardrails are a sensible checklist. Guardrail 5 calls for [meaningful human oversight](https://www.industry.gov.au/publications/voluntary-ai-safety-standard/10-guardrails), and guardrail 6 asks organisations to disclose when they use AI and when content is AI-generated. For a retailer, that translates into a person approving every AI-assisted image and a clear internal rule on when AI use is disclosed.

## A safe workflow for AI product images

1. **Start from a real photo of the actual product.** Use AI on the scene, not the subject.
2. **Lock the product pixels.** Background replacement and relighting are lower risk; generative "enhancement" of the product itself is high risk.
3. **Check scale and contents.** Does the product look the right size in the scene? Do any props look like they are included?
4. **Check against channel rules.** Main-image rules, text rules and AI metadata requirements differ by marketplace. See our [marketplace image requirements comparison](https://ecomchannels.com/insights/marketplace-image-requirements-compared/).
5. **Preserve or add required metadata.** Keep IPTC AI tags where Google requires them; add Amazon's synthetic-performer tag for AI-generated people.
6. **Human approval before publishing.** Someone who knows the product signs off.
7. **Keep records.** Store the source photo, the prompt or edit, and who approved it, in case a customer or regulator asks.

For guidance on where lifestyle scenes add the most value, see [lifestyle vs white background product photos](https://ecomchannels.com/insights/lifestyle-vs-white-background-product-photos/).

## Key figures

- Almost 90% of consumers globally want to know whether an image was created using AI ([Getty Images, 2024](https://newsroom.gettyimages.com/en/getty-images/nearly-90-of-consumers-want-transparency-on-ai-images-finds-getty-images-report)).
- 98% of consumers say authentic images and videos are pivotal to trust ([Getty Images, 2024](https://newsroom.gettyimages.com/en/getty-images/nearly-90-of-consumers-want-transparency-on-ai-images-finds-getty-images-report)).
- 45% of shoppers returned an item due to incorrect product content ([Salsify, 2024](https://www.salsify.com/hubfs/2024/Content/Salsify%20Consumer%20Research%20Report%202024.pdf)).
- $14 million in penalties for misleading water-resistance ads featuring phones in pools and the sea ([ACCC](https://accc.gov.au/consumers/advertising-and-promotions/false-or-misleading-claims), Federal Court order, 2022).

## How eComChannels helps

AI imagery is safest when the rules are built into the workflow rather than left to memory.

[eComChannels AI graphics](https://ecomchannels.com/products/ai-graphics/) is trained on each marketplace's image rules and builds image sets per channel from your real product photos. Every set goes to a person for approval before it publishes, so AI drafts and people decide, which matches the human-oversight approach in Australia's AI guidance. Approve once and the compliant sets publish to every connected marketplace, giving you fuller, more consistent galleries (better for visibility and conversion) without a studio day for every seasonal refresh.

[Book a demo or start a 30-day free trial](https://ecomchannels.com/book-a-demo/).

## FAQ

### Is it legal to use AI-generated product images in Australia?

Using AI is not in itself unlawful, but the images must not mislead. Under the Australian Consumer Law, images of what you offer must be accurate, regardless of how they were made or whether you intended to mislead. This is general information, not legal advice.

### Do I have to label AI-generated product images?

It depends on the channel. Google Merchant Center requires AI-generated images to carry AI metadata, and Amazon requires a metadata tag for photorealistic AI-generated people. Australia's voluntary AI guidance also recommends disclosing AI-generated content.

### What's the lowest-risk use of AI for product photos?

Background removal and replacement, cropping, resizing and lighting correction on a genuine photo of the product, with a person checking the result. Avoid generating or altering the product itself.

## Sources
- False or misleading claims, ACCC (2026): https://accc.gov.au/consumers/advertising-and-promotions/false-or-misleading-claims
- Product image guide, Amazon Seller Central (2026): https://sellercentral.amazon.com/help/hub/reference/external/G1881?locale=en-US
- Image link [image_link], Google Merchant Center Help (2026): https://support.google.com/merchants/answer/6324350
- About Product Studio for Merchant Center, Google Merchant Center Help (2026): https://support.google.com/merchants/answer/13708167?hl=en
- Photo Tips, eBay Australia Seller Centre (2026): https://www.ebay.com.au/sellercentre/create-a-great-listing/photo-tips
- Nearly 90% of Consumers Want Transparency on AI Images finds Getty Images Report, Getty Images (2024): https://newsroom.gettyimages.com/en/getty-images/nearly-90-of-consumers-want-transparency-on-ai-images-finds-getty-images-report
- The 10 guardrails (Voluntary AI Safety Standard), Department of Industry, Science and Resources (2025): https://www.industry.gov.au/publications/voluntary-ai-safety-standard/10-guardrails
- 2024 Consumer Research Report, Salsify (2024): https://www.salsify.com/hubfs/2024/Content/Salsify%20Consumer%20Research%20Report%202024.pdf