Google's new AI visual policy is a mandatory set of disclosure and quality rules that govern how AI-generated and AI-modified product imagery can appear in Google Search, Shopping, and Performance Max results. This matters for ecommerce sellers because the policy takes effect in 14 days and any catalog image that fails the new standards risks being suppressed, demoted, or rejected entirely from Google's commercial surfaces.
For brands publishing hundreds or thousands of SKUs, the policy is a search visibility deadline, not a creative suggestion. Treat it like a payment-processor compliance change, not a marketing refresh.
What the policy actually requires
Google's update, published in the Google Search Central blog and detailed in the Merchant Center documentation, introduces three binding rules for any image used in a Shopping listing or Performance Max asset group:
- AI-generated or AI-modified product imagery must be labeled in the feed using the new
image_ai_generatedattribute. - Any synthetic content that could mislead a shopper about material, scale, color, or function is now classified as a policy violation, not a soft warning.
- Image quality scoring now includes an "authenticity" sub-score that weighs lighting consistency, geometry preservation, and unedited texture detail.
The short rollout window is intentional. Google has stated that the 14-day window is meant to force immediate remediation rather than allow gradual compliance, according to Search Engine Journal's coverage of the announcement. Sellers who treat this as a quarterly project will see measurable listing suppression within the first crawl cycle after the policy date.
Why ecommerce is in the crosshairs
Ecommerce is the first vertical where the policy carries real revenue consequences. The new authenticity score feeds directly into the same quality multiplier that determines whether a product surfaces for high-intent queries. A single flagged image can drop a listing below the visibility floor for thousands of commercial keywords, and the recovery curve is slow once suppression begins.
According to a Statista analysis of Google Shopping click share, roughly 68% of product clicks on Google originate from Shopping panels, image packs, and Performance Max asset groups. That single number explains why a visual authenticity rule is more disruptive to ecommerce than to publishers or local businesses.
There is also a quality risk that pre-dates AI. Shopify's commerce research found that listings with five or more high-resolution images convert at 2.3 times the rate of single-image listings, which means the average catalog already leans heavily on visual supplements. If those supplements are AI-generated mockups, the new policy forces sellers to either disclose them or rebuild them with authentic photography before the deadline.
The policy is not banning AI imagery. It is banning undisclosed AI imagery that changes what the buyer sees. Honesty is the only compliance path that preserves ranking.
What counts as "AI-modified" under the new rules
The policy draws a sharp line between two categories. AI-generated imagery means any image where the pixels were synthesized from a model, including full AI background replacement that introduces new environmental context. AI-modified imagery means a real photograph that has been altered by a model in a way that changes material, geometry, or scale.
Critical detail: even with disclosure, modifications that change a buyer's perception of the product are disallowed. Adding a soft shadow is allowed. Replacing a matte ceramic texture with a glossy one is not. Swapping a fabric weave is not. This is the rule sellers most often miss in early guidance, and it is the one most likely to trigger a manual review under the new framework.
image_ai_generated attribute if the model generated any pixels. When in doubt, treat any AI-touched file as synthetic and disclose it.The 14-day action plan
Most mid-size catalogs have between 500 and 50,000 product images. A full reshoot is impossible in 14 days. A pure compliance flagging is possible but exposes the brand to a longer-term quality penalty. The fastest path is a triage workflow that separates the catalog into three buckets: high-value SKUs that need authentic photography, mid-value SKUs that can use compliant mockups, and long-tail SKUs that need disclosed AI imagery at minimum.
image_ai_generated attribute through the Merchant Center feed. Validate in the Merchant Center diagnostics panel before the policy date to catch rejected rows early and avoid last-minute feed errors.14-day compliance checklist
- ✅ Audit every catalog image and tag it authentic, AI-modified, or AI-generated.
- ✅ Identify your top 200 SKUs by revenue or impression share.
- ✅ Rebuild top 200 SKUs with authentic photography or a compliant mockup.
- ✅ Push the
image_ai_generatedattribute through your Merchant Center feed. - ✅ Run a diagnostic pass in Merchant Center before the policy date.
- ✅ Backfill lifestyle imagery for the long tail with a real-product-anchored tool.
- ✅ Document the policy version and audit date for internal compliance records.
Rewarx vs. generic AI image tools
Not every AI image tool is built for the new compliance environment. The table below compares what matters for a Google-compliant catalog.
| Capability | Rewarx | Generic AI image generators |
|---|---|---|
| Real product used as base | Yes — keeps original geometry, texture, and material | No — fully synthetic output |
| Background replacement without resynthesizing product | Yes — pixel-preserving on the product | Often re-renders the subject |
| Disclosed AI label feed-ready | Yes — exports image_ai_generated tag | No — manual mapping required |
| Compliant with Google authenticity score | Yes — designed for Shopping feeds | Varies — often fails the score |
The difference is structural. Tools that generate imagery from a text prompt ship a synthetic file. Tools that anchor on the real product and only edit the surroundings ship an authentic file that Google rewards. For ecommerce sellers, that distinction is the difference between a suppressed listing and a top-three Shopping placement.
Frequently asked questions
Does the new policy ban AI product images entirely?
No. The policy bans undisclosed AI imagery and any AI modification that changes the buyer's perception of the product. Disclosed AI imagery that preserves real material, geometry, and scale remains allowed across Shopping, Performance Max, and free listings.
What happens if my catalog is not compliant in 14 days?
Non-compliant images are scored into the lower quality band and progressively suppressed from Shopping surfaces. In practice, sellers report impression drops of 30% to 60% within the first full crawl cycle after the policy date, with manual review triggered for repeated violations.
Is a mockup image considered AI-generated under the new rules?
It depends on how the mockup is produced. If the product pixels come from a real photograph and the surrounding context is enhanced or composited, the image is typically classified as AI-modified and requires the disclosure attribute. If the product itself is synthesized, the image is fully AI-generated and is subject to the stricter rule about preserving buyer perception.
How do I add the disclosure attribute to my Merchant Center feed?
Add the image_ai_generated attribute to each product in your feed, set to yes for any image that contains AI-generated or AI-modified pixels, and submit the feed. Google will surface feed diagnostics in the Merchant Center within 24 to 48 hours.
Ship a Google-compliant catalog in 14 days
Rewarx rebuilds product imagery with real-product anchors, exports feed-ready disclosure tags, and keeps your Shopping quality score intact.
Try Rewarx Free