Google published a consolidated rulebook for image assets in Search ads that gates the format behind five account-level conditions, sets hard rejection criteria for eight categories of creative defect, and carves out an exemption allowing AI disclosure labels to sit on top of imagery that the same policy otherwise forbids.

The image assets format requirements page, part of the Google Advertising Policies Help Center, was updated to incorporate changes flagged in a July 9, 2026 changelog entry titled "Updates to AI labeling requirements (July 2026)." Two documents now operate together. One sets the aesthetic and technical floor for images attached to text ads. The other punches a deliberate hole in it.

That combination is the substantive development. Advertisers running Search campaigns have worked under a text overlay prohibition for years, a rule designed to keep image assets showing products rather than promotional graphics. The AI labeling regime arriving across the European Union, India and New York requires the opposite: visible text on creative, telling consumers that AI was involved. Google has now written the exception into the policy rather than leaving advertisers to guess whether a compliance label counts as a policy violation.

Who can use image assets at all

The eligibility conditions are the least discussed part of the document and the most consequential for newer accounts. According to Google, an advertiser should be able to use image assets once the account has generated sufficient history on the platform and satisfies five separate conditions.

The account must have been open for at least 60 days. It must have what the documentation calls a good history of policy compliance. It must have active campaigns, and it must have active text ads with Search spend in the last 30 days greater than zero. Finally, the account must sit in an eligible vertical or sub-vertical.

That last condition is a categorical exclusion rather than a threshold. Google names sexual content and gambling as examples of sensitive verticals or sub-verticals that are not eligible for image assets. No spend level or compliance record unlocks the format for accounts in those categories.

The 60-day requirement combined with the 30-day spend window means a newly launched account cannot attach images to its Search ads for roughly two months, regardless of budget. For agencies onboarding clients onto fresh accounts, or brands migrating between account structures, the format is unavailable during the period when campaign structure is typically being established.

The specifications

Two file formats are accepted: PNG and JPG. The maximum file size is 5120 KB.

Square imagery at a 1x1 ratio is required. It serves on Google.com and on YouTube search via AdSense For Search, the Search Partners inventory. Landscape at 1.91x1 is optional but recommended, and serves on the same surfaces.

Minimum resolution for square assets is 300 by 300 pixels, with 1200 by 1200 recommended. Landscape assets require at least 600 by 314 pixels, with 1200 by 628 recommended. Those landscape minimums match the figures Google has published for Demand Gen campaigns, where the same 600 by 314 landscape and 300 by 300 square floors apply alongside an identical 5120 KB file ceiling.

The recommended safe area places the most important content in the center 80 percent of the image. Google notes that images can be cropped using the image picker during implementation.

Eight ways an image gets rejected

The requirements enumerate specific defect categories, each with allowed and not-allowed examples.

Text or graphic overlay covers images carrying text or graphics, including brand logos or brand logos used as the image asset itself. This category is allowed with limitations. For Performance Max campaigns, overlays are permitted, though Google recommends supplying at least one image without overlays for each aspect ratio: square, horizontal and vertical.

Excessive blank space applies where whitespace dominates or where the product or service offering cannot be recognized because of how the subject is proportioned within the frame.

Collage targets images combined together in post production. The limitation carve-out covers naturally occurring collages, which Google lists as including scrapbooks, greeting cards, printing catalogues and yearbooks. The distinction is whether the composite existed as a physical object or was assembled in an editor.

Blurry or unclear images that are unrecognizable are not allowed. Distorted images covers creative visually skewed or warped in a way that makes the subject difficult to understand. Poorly cropped images addresses cropping that makes the subject difficult to distinguish. These three quality criteria echo standards Google applied to YouTube and Discover Feed assets in October 2025, where blurred, distorted or poorly cropped images similarly failed editorial review.

Prohibited content covers nudity, including nudity used for artistic purposes, and sexually suggestive images. Two limitations apply. Modeled clothing is allowed where there is good coverage and the model is not showing excessive skin. Lingerie and bathing suits are allowed when pictured as a flat lay or on a mannequin.

A stricter standard than the rest of Google

One passage in the requirements deserves attention from anyone managing creative across multiple Google surfaces. According to Google, image assets must comply with all Google Ads and personalization policies, and each asset is additionally reviewed against the image asset requirements, which take priority where a conflict arises.

The documentation then states plainly that the image asset requirements are more restrictive than the ads requirements for other platforms and surfaces, meaning disapproved image assets may continue to run on Google's other owned and operated properties.

An image rejected as a Search image asset is therefore not necessarily rejected everywhere. The same file may serve without issue on other Google inventory. Disapproval in one surface carries no automatic implication for another, which complicates asset library management for teams treating a single approved creative pool as portable across campaign types.

The pinning problem

A technical note buried in the requirements has direct consequences for regulated advertisers. Headlines or descriptions pinned to Headline position 1, Headline position 2 or Description position 1 will always show. Description position 2 may be truncated when an image asset is displayed.

According to Google, advertisers with legally required text that must appear in every ad must pin it to Headline position 1, Headline position 2 or Description position 1. The instruction appears twice in the document, once as a note within the eligibility section and again as a tip alongside the AI labeling material.

For financial services, pharmaceutical, insurance and other advertisers carrying mandatory disclaimers, this converts a creative optimization decision into a compliance constraint. Pinning reduces the combinatorial space available to responsive search ad optimization. Google's own guidance on responsive search ads sets asset learning thresholds at more than 500 impressions for individual assets and over 2,000 impressions for complete ads in the Google Search Top segment across a 30-day period. Fewer rotating combinations means those thresholds are reached across a narrower asset set.

The AI label exemption

The July 9 changelog is where the two documents connect. According to Google, starting in July of 2026 the company permits adding text or visual labels directly within image and video ad creatives that were generated or modified using AI, to help advertisers comply with emerging AI transparency regulations.

Google Ads may also automatically apply labels to certain assets generated by Google AI tools. The changelog states that these labels will not be in violation of Google policies prohibiting text overlays and watermarks.

The regulatory driver is named directly. AI regulations in the European Union, India and New York require that ads with certain AI-generated or edited assets include disclosures or labels informing consumers that the ads were made with AI.

Advertisers have two routes. They can add labels directly to their creatives, or use the AI label setting, which the changelog says will launch gradually throughout July across five products: Google Ads, Display & Video 360, Campaign Manager 360, Merchant Center and Ads Editor. That five-product rollout was documented in PPC Land's coverage of the AI label setting, which detailed the interface controls including a bulksheet column in Display & Video 360 and bulk labeling through the shared asset library.

The disclaimer attached to the setting is unambiguous. Use of the AI label setting in Google's advertising products does not guarantee compliance with specific regulations. The documentation directs advertisers to seek legal guidance and take measures as needed to ensure their ads and assets follow all local legal obligations and policies.

PPC Land has reported that this construction places the compliance obligation on advertisers rather than on Google or on publishers. The advertiser selects, or fails to select, the label. Google supplies an interface and no warranty.

Election advertisers face a separate and unchanged obligation. According to the changelog, they should continue to disclose all election ads containing synthetic or digitally altered content by selecting the checkbox in the Altered or synthetic content section of campaign settings, as outlined in the political content policy. That requirement predates the July 2026 update and operates independently of the AI label setting.

Why the overlay carve-out matters operationally

Consider the sequence a creative team now faces. An image asset must show the product clearly, avoid overlays, avoid collage assembly, avoid excessive whitespace, and keep important content within the center 80 percent of the frame. If that asset was generated or edited with AI and the campaign targets the European Union, India or New York, it must also carry a visible disclosure.

Those two sets of instructions pull against each other, and Google has resolved the conflict by exempting the disclosure from the overlay rule. What the documentation does not specify is how much text a disclosure may contain before it stops functioning as a disclosure and starts functioning as an overlay. The exemption is stated as a category rather than as a measurable threshold.

Prior PPC Land reporting on the labeling system noted Google guidance that a custom label should sit outside a margin of roughly 5.5 percent around an asset's full perimeter, since ad rendering can trim the outer edge. That figure and the center 80 percent safe area in the image requirements describe the same underlying problem from opposite directions: the edges of an image asset are not reliable real estate, yet a compliance label must be visible.

Regulatory timing

The August 2, 2026 date for Article 50 transparency obligations under the EU AI Act sits three weeks after the labeling changelog was posted. Brussels negotiations over the Digital Omnibus package collapsed in May 2026 without producing a delay to that date, according to PPC Land coverage of Google Ads API v24.2, which documented the developer-facing synthetic content labeling structures Google released on June 24, 2026.

The API release introduced attestation across two dimensions: whether assets and ads are AI-generated, and whether content was generated fully automatically or with advertiser review. That second distinction, fully automated versus human-in-the-loop, maps onto the transparency requirements the AI Act introduces.

More recently, the European Commission published Article 50 implementation guidelines and a finalised Code of Practice on Transparency of AI-Generated Content, with penalties reaching 3 percent of turnover for non-compliance. The Code sets technical thresholds including watermarking obligations for free-form text longer than 200 tokens.

Context for advertisers and agencies

The image assets page is not new. What changed is that it now carries the AI labeling exception, making it the single document a Search advertiser consults for both creative quality standards and AI disclosure permissions.

The disclaimer at the foot of the eligibility section also warrants notice. By opting into image assets, an advertiser confirms and acknowledges two things: that the advertiser owns all legal rights to publish the images and to share them with Google, and that the advertiser instructs Google to publish the shared images on its behalf for advertising purposes.

That rights attestation intersects with AI-generated creative in ways the documentation does not address. An advertiser using a third-party generative tool is attesting to ownership of an output whose provenance may be contested. Google's help documentation on the labeling side has stated that not all AI-created content needs to be labeled, with requirements varying by region and content type, which means the same asset may carry a label in one market and none in another while the ownership attestation remains constant.

Image assets sit within a broader pattern of Google tightening asset-level policy across formats. In October 2025, message assets became subject to a verification requirement, with unverified assets disapproved and prevented from serving. In November 2024, Google restricted standalone buttons in image ads as a misleading design practice, effective January 23, 2025.

The direction is consistent: more specificity at the asset level, more conditions on eligibility, and more of the compliance burden expressed as advertiser attestation rather than platform verification.

Timeline

Summary

Who: Google, through its Advertising Policies Help Center, addressing advertisers and agencies running Search campaigns with image assets, and specifically those operating in the European Union, India and New York.

What: An updated image assets format requirements page setting five account eligibility conditions, two accepted file formats, a 5120 KB size ceiling, minimum resolutions of 300 by 300 pixels for square and 600 by 314 for landscape, and eight defect categories that trigger disapproval. The page now incorporates an exemption allowing AI disclosure labels to appear on creative without violating Google's text overlay and watermark prohibitions.

When: The changelog entry was posted on July 9, 2026. The AI label setting rolls out gradually throughout July 2026. Article 50 transparency obligations under the EU AI Act take effect on August 2, 2026.

Where: Image assets serve on Google.com and on YouTube search via AdSense For Search, the Search Partners network. The AI label setting spans Google Ads, Display & Video 360, Campaign Manager 360, Merchant Center and Ads Editor.

Why: AI regulations in the European Union, India and New York require disclosure when ads contain certain AI-generated or AI-edited assets. Google's existing image asset policy prohibited text overlays, creating a direct conflict between platform rules and legal obligation. The exemption resolves that conflict, while Google states that use of the label setting does not guarantee regulatory compliance, leaving the obligation with the advertiser.