Google Ads Editor 2.13 Adds a Field Where You Declare AI-Generated Creative

AI disclosure just became a bulk-edit field

A performance marketer opens Google Ads Editor on a Monday, selects four hundred video assets that a small team generated the week before, and sets one dropdown for all of them. The dropdown is new. It is called User attestation, and it is where you tell Google whether the creative was made with AI.

Google's release notes for Ads Editor version 2.13 describe it plainly: a new compliance field titled User attestation, available for Image and Video assets to manage disclosure mandates for AI-generated content. That is the whole feature. No detection model, no scan of your file, no verdict handed down. A field, three states, and the option to apply it across a selection.

The headline on the release went to something else. Search Engine Land reported on July 24, 2026 that 2.13 brings AI Max support to Shopping campaigns, which is true and is the bigger line item for most search teams. But if you make ads with AI, the attestation field is the one that changes your week.

What the User attestation field actually does

Three states, and the interesting one is Unset

Google lists the options as Unset, Label, and Don't label. Two of those are answers. The third is the absence of an answer, and it is almost certainly the default that most accounts will sit in for a long time.

• Unset: no declaration recorded. The starting position for every existing asset in your account.
• Label: you are telling Google the asset carries AI-generated content.
• Don't label: you are telling Google it does not.

That matters because a three-state design is what you build when the requirement is not yet universal. A binary yes-or-no forces every advertiser in every market to answer today. Unset lets Google collect the declaration where a market or a policy already demands it, while leaving the rest of the world untouched until it does not. Read the field as infrastructure being laid ahead of the rules it will serve.

It also means silence is currently a valid state, which will not feel true forever. Fields that begin with a null option tend to lose it once the surrounding policy hardens, and the accounts that get caught out are the ones that treated the null as a decision rather than as a deadline they had not met yet.

It applies in bulk, not asset by asset

Google's notes say the settings can be applied in bulk or directly within asset pickers. That single clause is the tell. Ads Editor exists to do things to hundreds of rows at once, and building the attestation into that surface means Google expects the answer to be uniform across large batches of creative rather than considered one file at a time.

Which is a reasonable expectation. An advertiser who generates video with an AI tool does not generate some of it with AI. The whole library carries the same answer, so the field that carries the answer should accept the whole library in one action.

It is a declaration, not a detector

Nothing on the page describes analysis, watermark reading, or provenance checking. The field records what you say about your own asset. That puts the burden of accuracy on your production records, not on Google's ability to spot a generated frame.

We covered the viewer-facing half of this earlier: Google's How this ad was made panel is what a person sees when they tap through on an ad. The attestation field is the other end of the same pipe, the place where the advertiser supplies the input. Seeing both ends now exist is more informative than either one alone.

Why a dropdown is a bigger signal than an announcement

Policy pages ask. Fields collect.

Google has published AI-disclosure expectations before. Its election-ads policy already states that advertisers must disclose all election ads that contain synthetic or digitally altered content, and we walked through the wider picture in the 2026 AI ad label rules and what actually counts as an AI-generated ad.

A policy page is a rule you are expected to read. A field in the tool you use every day is a rule that arrives at your desk. The distance between those two is where most compliance actually lives, and 2.13 closed it for AI disclosure.

Bulk tooling is built for people running volume

Ads Editor is not the casual surface. It is the desktop application agencies and in-house teams use to push large structural changes into accounts offline. Shipping the attestation field there first, with bulk apply, says something about who Google thinks needs it: advertisers with enough AI-made creative that answering per asset would be unworkable.

That is the profile of anyone doing serious creative testing right now. If your process is to ship a batch of variants and let the data pick the winners, you are exactly the account this field was designed around.

The rule is not the interesting part, the plumbing is

Plenty of teams are still waiting to see whether AI disclosure gets enforced hard, softly, or unevenly by market. That debate will run for a while. Meanwhile the plumbing is being installed, and plumbing is harder to reverse than a policy blog post.

Once a field exists, it accumulates. Reporting gets built on it, filters get built on it, and eventually a policy references it by name. The unglamorous version of a rule taking effect is a dropdown appearing in a release note.

AI Max arrives in Shopping campaigns, with the controls attached

What Editor 2.13 exposes

Google's release notes state that a comprehensive suite of AI Max-related capabilities is now available for Shopping campaigns, achieving feature parity with Search and Performance Max campaigns. The specifics listed are automated text, URL expansion controls, text guidelines, URL exclusions, and brand list controls.

Parity is the operative word. AI Max controls that search teams already had in Search and Performance Max now exist for Shopping, in the offline bulk tool, which is where a retailer with thousands of SKUs would want to set them.

That location matters more than it sounds. A control that only lives in the web interface gets configured by whoever happens to be in the account that day. A control in Editor gets configured once, in a file, and pushed to every campaign at the same time, which is how a large retailer keeps a brand rule from drifting across a catalog.

The controls are mostly about text, which means copy

Look at the list again. Automated text, text guidelines, URL expansion, URL exclusions, brand lists. Four of those five are ways to fence in language: what the system may write, where it may send the click, and which brand terms it may use.

Google's own AI Max for Shopping page describes the text customization piece as generating ad copy for your Shopping ads so that they speak directly to shopper intent and conversational queries. So the platform writes the copy, and the controls in Editor are how you constrain what it writes. That is the same trade every advertiser is now making across formats, and it is the reason Google's creative standards are worth reading before you hand over the copy layer.

Available in Editor is not the same as switched on in your account

One caution, and it is the reason to cite the changelog rather than the marketing page. Two Google surfaces currently disagree with each other:

• The Editor 2.13 release notes say the AI Max capabilities are now available for Shopping campaigns.
• Google's AI Max for Shopping announcement page still says the one-click toggle is coming to Shopping campaigns, in the future tense.

The changelog is the one that describes what a build actually ships, so that is the surface to cite. The announcement page is an earlier marketing post that was not refreshed as the rollout moved.

The honest reading: 2.13 gives you the controls in the tool. Whether AI Max is enabled for your Shopping campaigns is a separate rollout question, and you should check your own account rather than assume the release note settled it.

Demand Gen keeps absorbing the rest of the stack

Maps becomes a selectable channel

You can now select and enable Maps as a channel option within the Channels settings panel for Demand Gen ad groups. For local and multi-location retailers that is a new placement lever available in bulk, which is the only way it is usable at scale.

Classic display assets move in

Support for Classic display assets has been added to Demand Gen image ads. Google frames it as using standard image dimensions from plain Image ads to streamline migrating legacy Display campaigns into the Demand Gen format. Read the word migration carefully. This is a bridge built for traffic moving one direction.

Generated text gets an on and off switch

You can now opt in or opt out of the Generated text enhancement specifically for Demand Gen video ads, which Google describes as direct control over automated copy variations. This is worth separating from the attestation field, because they answer opposite questions.

• Attestation asks: did you generate this asset?
• Generated text asks: may Google generate copy on top of it?

A team can honestly answer Don't label on a hand-shot video while the platform is writing headline variations for it. Two different kinds of machine involvement, two different switches, and only one of them is a disclosure.

Creating new Video Action campaigns is gone

Under deprecated features, Google states that support for creating new Video Action campaigns has been removed, that active campaigns have been migrated to Demand Gen, and that creation restrictions apply across standard accounts.

For most readers this is a footnote. For anyone whose bulk workflow builds Video Action campaigns from a spreadsheet template, it is the line item that breaks something on Monday. Demand Gen is now the destination for that video demand, which is consistent with the direction we traced in Google's video campaign groups rollout.

The rest of the release, in one pass

The remaining items are smaller, but two of them are worth a second look if you run video.

Item	What it changes
Business name and logo	Dedicated fields on Responsive Video and Non-skippable video ads
Customer retention goal	Reengagement goal support for Performance Max
Channel performance reports	Native channel breakdown inside Editor
Column management	Presets, reordering, CSV import and export
Resume download	Re-select the same campaigns after a failed download

The branding fields are the quiet upgrade for video advertisers. Dedicated Business name and Business logo fields now sit in the properties panel for both Responsive Video ads and Non-skippable video ads, which Google says gives consistent visibility of those assets across video formats. If your brand mark previously depended on whatever was burned into the creative, it no longer has to.

Channel performance reporting is the workflow win. The native reports replicate the channel performance breakdown from the Google Ads web interface, so a bulk operator can read channel metrics without leaving Editor for the browser.

What to do if your creative is AI-made

Decide a default attestation posture, and write it down

Do not answer this field asset by asset under time pressure. Decide once, at the team level, what your standing answer is for each source of creative, then apply it in bulk the way the tool intends.

Most teams will land somewhere simple:

• Generated in an AI video or image tool: Label.
• Shot on a phone or a camera by a real person, unedited by generative tools: Don't label.
• Shot by a person, then retouched or extended with generative tools: Label, because the asset now contains generated content.
• Origin genuinely unknown: Unset, with a note to resolve it rather than a shrug.

The value is not the specific policy, it is that the policy exists before the deadline does.

Keep the production record at the source

The field records your claim. Your claim is only as good as your memory of where the file came from, and six months of creative testing destroys that memory fast.

The fix is boring and effective:

• Keep AI-generated and camera-shot assets in separate places from the moment they are produced.
• Name files so the origin survives a download into someone else's machine.
• Never merge the two folders to save space, no matter how tempting at cleanup time.
• Record the tool, not just the fact of AI, so you can answer a follow-up question later.

If the origin is legible from the filename, a bulk attestation takes seconds. If it is not, you are guessing, and guessing is the failure mode a disclosure field is least forgiving of.

Run the audit once, before anyone asks

There is a version of this that takes an afternoon and a version that takes a quarter. The short one is worth doing now, while the field is new and there is time to get the answer right without a deadline attached.

Pull your active image and video assets, sort them by the folder or tool they came from, and mark the ones you cannot place. That last pile is the real output of the exercise. It tells you how much of your library has already lost its provenance, and it is the only number in this whole release that is specific to your account rather than to Google's roadmap.

Do not treat disclosure as a performance problem

The reflex is to assume a label costs you conversions, so the best move is to avoid triggering one. That reflex is worth dropping. Disclosure has been moving the same direction across every major platform, from Google's transparency panel to TikTok's AI disclosure rules and Google's own rules for AI-generated images in ads.

What actually protects performance is creative that earns trust on its own terms. An ad that reads like a real person's honest take about a product survives a disclosure line. An ad that only worked because nobody knew how it was made was always fragile.

How Novoads solves the attestation bookkeeping problem

If every ad in a library came from the same place, the attestation answer is never ambiguous. That is the practical shape of working in Novoads: you upload a product image or write a script, pick an AI actor, and get an ad-ready video, so the origin of every asset is known by construction rather than reconstructed later from a folder name.

That makes the new field a one-time decision instead of an audit. Set your posture once, apply it in bulk, and spend the time you saved on the creative itself. Access starts at $1 for 3 days and then continues on the $49 per month Inicial plan, and you can cancel whenever you want.

Disclosure is becoming a field, not a debate

The industry has spent two years arguing about whether AI-made advertising should be labeled. While that argument continues, the answer is quietly being turned into a dropdown with three states, a bulk-apply button, and a place in the release notes between a bidding change and a column manager.

That is what a rule looks like right before it stops being interesting. The teams who will find this easy are the ones who already know, without checking, which of their assets a machine made.