What Google actually announced on July 9
Keerat Sharma published "Expanding AI transparency in ads" on the Google Ads blog on July 9, 2026. The announcement introduced a two-tier disclosure system and extended AI labeling to Search, YouTube, and Discover ad formats. Search Engine Land confirmed coverage the same day.
The consumer-facing touchpoint is the My Ad Center panel. Users who tap the three-dot or information icon on an ad will see a "How this ad was made" section that discloses whether AI was used in production. This panel already existed for other transparency purposes; the July update adds AI provenance as a distinct disclosure layer.
The two tiers work as follows:
- Tier 1 (automatic): Ads created using Google's own generative AI advertising tools receive AI disclosure automatically. No advertiser action is required.
- Tier 2 (manual): Ads created or edited using AI tools outside Google require the advertiser to use a new manual disclosure control within the Google Ads interface before the ad runs.
Key distinction
The manual disclosure control is a new operational step, not a setting you toggle once. It applies per-creative, meaning every AI-assisted asset uploaded from a third-party tool needs the control activated before submission.
Google had a precedent for this: election ads in certain markets had required disclosure of synthetic or AI-altered content since 2023. The July 2026 update extends that logic to all advertising categories.
The scope surprise: "created or edited with AI"
The phrase Google used is "created or edited with AI." Most practitioners read "AI-generated" and assume it means fully synthesised images, voices, or copy. The "or edited" clause changes the calculation significantly.
Common AI editing actions that trigger the disclosure requirement include:
- Background removal or replacement using AI tools (Adobe Firefly, Canva AI, Remove.bg)
- Image extension or outpainting to change aspect ratios
- AI-generated voiceover or voice cloning for video ads
- Copy refinement or rewriting using ChatGPT, Claude, Gemini, or Jasper before paste into Google Ads
- Object insertion or removal using generative fill
- Colour grading or relighting applied via AI tools
If a human photographer shot the original image and an AI tool then extended the frame to fit a YouTube format, that creative requires manual disclosure under the current policy wording.
| Action | Tool examples | Disclosure required? |
|---|---|---|
| AI background removal | Adobe Firefly, Canva AI, Remove.bg | Yes (editing) |
| AI image extension / outpainting | Adobe Firefly, Midjourney, DALL-E | Yes (editing) |
| AI copy generation or rewrite | ChatGPT, Jasper, Claude, Gemini | Yes (created) |
| AI voiceover / voice cloning | ElevenLabs, Murf, HeyGen | Yes (created) |
| AI video generation or editing | Runway, Pika, Kling | Yes (created/edited) |
| Google Performance Max auto-creative | Google Ads (native) | Automatic (no action) |
| Google AI-generated images in asset library | Google Ads (native) | Automatic (no action) |
| Manual crop or resize (non-AI tool) | Photoshop, Sketch | No |
Watch out
Teams that use AI tools in the middle of a human creative workflow (not for full generation) are most at risk of missing this. The final image may look entirely human-shot while containing AI-edited elements that trigger the requirement.
Which formats and markets are in scope
The July 9 announcement explicitly names Search, YouTube, and Discover as the formats in scope. Display, Shopping, Performance Max, and App campaigns are not listed in the announcement. Whether the policy extends to those formats in a subsequent update is not yet confirmed.
| Format | In scope (July 9, 2026) | Notes |
|---|---|---|
| Search | Yes | Responsive search ads, expanded text ad variants |
| YouTube | Yes | In-stream, bumper, non-skippable, Shorts |
| Discover | Yes | Image and text creative in Discover feed |
| Display | Not listed | Monitor for updates |
| Shopping | Not listed | Monitor for updates |
| Performance Max | Not listed | Google-native assets already automatic; third-party uploads unclear |
| App campaigns | Not listed | Monitor for updates |
On market scope, Google's announcement applies globally but notes that visible on-ad labels (as opposed to the My Ad Center panel) will appear "based on local requirements." This is the clause that will create regional divergence. Markets where regulators have issued specific AI disclosure requirements for advertising will likely see visible labels on the ad unit itself, not just in the My Ad Center panel. Advertisers running across Singapore, Australia, the US, Canada, and Malaysia should assume the disclosure system will behave differently in each market as local requirements clarify.
What your team needs to change right now
The operational change is straightforward but requires updating your creative QA process before campaigns go live, not after. The manual disclosure control in Google Ads is the step most teams will miss because it sits outside the existing creative handoff and copy approval workflow.
If you use only Google's own AI tools (AI-generated images in asset library, automatically created assets, Gemini copy suggestions natively in the interface), no action is required. Disclosure is automatic.
If you use any external AI tool in creative production for Search, YouTube, or Discover ads, follow this checklist before submission:
- Identify every asset in the creative set (headline copy, images, video, voiceover, background) and flag which were created or edited with an external AI tool.
- For each flagged asset, use the manual disclosure control in Google Ads to indicate AI use. This control should appear in the asset upload or campaign creation flow following the July 9 update.
- Document the AI tools used per asset in your creative brief or production record. This protects the advertiser if Google later audits compliance.
- Update your creative brief template to include a field: "AI tools used in production: [yes/no, tools used]" so the disclosure step is triggered at the creative handoff stage, not during campaign setup.
Practical note
The manual disclosure control is new as of July 9, 2026. If your Google Ads interface does not yet show the control, check that your account is on the latest interface rollout. Google typically stages UI releases over several weeks. If the control is not visible, note this in your compliance record and add the disclosure step as soon as it appears.
For agencies managing multiple advertiser accounts, the disclosure obligation sits with the advertiser account, not the agency. Make sure your client-facing creative handoff process includes confirmation of AI tool use so the account owner can apply the disclosure correctly.
The visible label risk you have not planned for
Most discussion of this update focuses on the My Ad Center panel, which is user-initiated (the user has to tap the icon to see the disclosure). That is the lower-stakes scenario. The higher-stakes scenario is the jurisdiction-dependent visible label that appears directly on the ad creative.
Google's announcement states: "based on local requirements, a label may also appear directly on the ad." Google has not published which markets activate this, what the label looks like, or where it is placed on the creative. This is the planning gap that most advertisers have not addressed.
The design implications if a visible label appears on the ad unit:
- Creative that fills the frame edge-to-edge may have a label overlaid on the main visual or headline, affecting legibility and brand presentation.
- Campaigns that rely heavily on precise visual composition (luxury goods, financial services with regulated copy) face the most disruption if a visible label changes the perceived framing of the ad.
- Advertisers cannot currently design around the label because Google has not published its dimensions or placement specification.
Risk item
Google released no creative specification for the visible on-ad label alongside this announcement. Until that specification is published, the safest approach for markets likely to require it is to keep critical visual and copy elements away from corners and edges of the ad frame where disclosure badges typically appear.
One structural note: Google's SynthID watermarking technology is already embedded in output from Google's AI advertising tools. This invisible watermark predates the disclosure system and operates beneath the visible layer. It does not change what advertisers need to do, but it explains why Google can technically audit whether a piece of creative originated from its own tools versus a third-party tool.
Where regulatory direction is heading
Google's move is consistent with the regulatory trajectory across all five of leapbuzz's priority markets. The platforms are self-regulating ahead of formal requirements, which typically gives advertisers a shorter runway than they expect once regulations formalise.
| Market | Current status | Direction |
|---|---|---|
| European Union | EU AI Act Code of Practice started November 2025; advertising sector in scope | Mandatory AI content disclosure for high-risk categories progressing through implementation |
| United States | FTC has established AI disclosure requirements for influencer testimonials and endorsements | Direction toward broader AI advertising disclosure; no comprehensive federal rule as of July 2026 |
| Singapore | PDPC and IMDA have issued AI governance frameworks with transparency principles | Sector-specific guidance expected; financial sector (MAS oversight) likely to see requirements first |
| Australia | ACCC monitoring digital platforms for AI content; Australian Consumer Law applies to misleading AI representations | Mandatory labeling framework under active discussion as of 2026 |
| Canada | Bill C-27 (Artificial Intelligence and Data Act) stalled; consumer protection guidance evolving | Disclosure direction consistent with international peers; timeline unclear |
The practical implication for multi-market advertisers: build the disclosure habit now, before local requirements formalise the specific format. Google's two-tier system gives you the mechanism. Use it consistently across all markets, not just those with active requirements, so that compliance is a workflow default rather than a market-by-market exception.
For sectors with higher regulatory sensitivity (financial services, healthcare, regulated consumer goods), the timeline between voluntary platform disclosure and formal regulatory mandate has historically been shorter than expected. Do not plan on a multi-year runway.