Google today told AdSense publishers that visible AI labels may begin to appear on ads served to their websites and apps in California, South Korea and Vietnam, extending to the publisher side a disclosure system that until now covered the European Union, India and New York. The notice, posted on the AdSense Help Center on November 12, 2026, says the change starts in November 2026 and applies only to creatives bought through Google Ads and Display & Video 360. Ads that come from other buyers, or that publishers sell and traffic themselves, get no automatic label from Google.

In Short

Google said today that ads bought through Google Ads and Display & Video 360 may carry a visible AI label when they run on publishers' sites and apps in California, South Korea and Vietnam, starting this month. The change matters to websites and apps that earn money through AdSense, Ad Manager or AdMob and reach people in those places, because the labels will appear on their pages while Google adds none to ads bought through other platforms or sold directly. For those other ads, adding AI disclosures falls to the buyers, their creative partners and the publishers who traffic them, using methods Google describes separately for display, video and native formats.

What the notice says

The announcement is short. Titled "Google's visible AI Labels expanding to new regions," it runs to two paragraphs. "In response to upcoming regulatory requirements in California, South Korea, and Vietnam, Google is extending its AI Labeling functionality in these regions starting November 2026," according to the notice. "For publishers, these AI labels may appear on creatives served to their properties when coming from Google demand (Google Ads, Display & Video 360)."

No specific start day is given. Nor does the notice name the laws it refers to. Its reference to Vietnam links to an English-language text of the country's Law on Artificial Intelligence published on the Vietnamese government's news portal. California and South Korea get no equivalent reference.

The second paragraph sends publishers to a longer help page, "Navigate AI labeling regulations," which sets out how Google's publisher products handle AI disclosure across Google Ad Manager, AdMob and AdSense. That page carries no publication date. It describes the regulatory backdrop in a single sentence: "These laws generally require certain AI-generated or edited content to be labeled, among other things."

Three kinds of demand, three different answers

The core of the guidance page is a table that splits ad demand into three sources. Each gets a different answer on who is responsible for disclosure.

The first is Google demand, meaning ads bought through Google Ads and Display & Video 360 and served through Ad Manager, AdSense or AdMob. For these, "Google buying platforms will offer assistive tools to help advertisers add appropriate AI labels to creatives," according to the table. These are the ads on which the new labels in California, South Korea and Vietnam may appear.

The second is third-party programmatic demand: Authorized Buyers, Open Bidding and SDK Bidding, again across all three publisher products. Here Google steps back. "Publishers should work with their third-party buying partners to understand their plans for compliance related to AI disclosures," the table says. A later section of the page is blunter still: "Google publisher products do not provide automated labeling for third-party creatives." It adds that "Third-party buyers are responsible for determining their compliance obligations and ensuring that their ad creatives contain appropriate disclosures."

The third is ads trafficked directly by the publisher, which the page lists as reservations, house ads and ads delivered through header bidding, among others. This row applies to Ad Manager and AdMob only, since AdSense publishers do not traffic their own campaigns. "Publishers should ensure that the creatives they traffic contain appropriate AI disclosures from their buying and creative partners," the page states.

The practical result is a split screen. On a single news site in Seoul or Los Angeles, an AI-assisted ad from a Google Ads campaign may arrive with a label already attached, while an equally synthetic creative bought through a third-party exchange via Open Bidding arrives without one, unless the buyer has added it. For AdSense, the second category matters more than it once did: since November 6, 2025, the product's blocking controls have been organised around Authorized Buyers, after Google replaced the old ad networks control with an Authorized Buyers list and switched new buyers to allowed by default.

Display, video and native: the technical routes

The guidance page then turns to formats. Its advice is directed at the parties that build and deliver creatives, but several passages place work on publishers as well.

Display

For display ads, the page says creative vendors and buyers "have a variety of mechanisms to update creative JavaScript to include AI disclosures when the creative renders." The alternative is simpler: display creative providers can alter the creative file itself so that the disclosure is part of the image.

Video

Video is where the publisher's technical setup starts to matter. According to the page, creative vendors and buyers can add the <Icons> element to the XML of a VAST response to render AI disclosures and handle the click-through. Google links to page 58 of the IAB Tech Lab's VAST 4.3 specification, where the element is defined. The Icons node has typically carried privacy overlays. Google is now pointing it at AI disclosure.

Where the publisher uses Google's Interactive Media Ads (IMA) SDK, rendering is automatic: "when present, the IMA SDK will handle the corresponding rendering and click through behavior automatically," according to the page. Publishers who rely instead on Publisher Authenticated Inventory (PAI) or the Programmatic Access Library (PAL) "will need to manually implement support" for rendering the icons and their click-through. That is engineering work on the publisher's player, and the page offers no timeline for it.

As with display, video creative providers can also edit the video file directly so that the disclosure is burned into the footage.

Native ads and custom badging

Native advertising works differently, because the publisher renders the ad's components in its own layout. Google's guidance points to the OpenRTB implementation it uses, which supports a "privacy" field for a "link to an ad preferences or opt-out page." Third-party demand can choose where that link leads, and the landing page "may contain additional AI disclosures," according to the page.

In other words, for third-party native ads, an AI disclosure may sit one click away from the ad rather than on it.

A separate note covers publishers that Google has allow-listed for custom ad transparency and badging, a group that designs its own ad markers. For them, the arrangement is reversed. "Google demand will not automatically include disclosures in this case, and will instead pass a field in the ad response to publishers notifying them that the native ad may include AI-generated content," the note says. These publishers are told to work with their account teams on best practices. The signal arrives; the rendering is theirs.

The page closes with a disclaimer that applies to all of the above: "This article is for informational purposes only and does not constitute legal advice."

A second notice in four months

Today's announcement is the second time this year that Google has addressed AI labels on the publisher side, and it follows changes made on the buying side first.

On July 9, 2026, Google added a "How this ad was made" panel to My Ad Center to disclose when generative AI was used to create or edit ads on Search, YouTube and Discover. The same day it introduced an AI label setting across Google Ads, Display & Video 360, Campaign Manager 360, Merchant Center and Ads Editor. Alongside the global panel, a visible overlay on the ad itself was reserved for campaigns targeting the European Union, India and New York. Google's documentation said it may apply labels without advertiser input in three cases: where Google itself is legally required to, where it receives signals from other platforms, and where an advertiser uses Google's fully automated creative features. Labels applied in those cases cannot be overridden.

Four days later, on July 13, an AdSense Help Center notice told publishers that Google's sell-side platforms offered no AI labeling features of their own, and that labels might appear on creatives from Google Ads or Display & Video 360. PPC Land reported at the time that Google had placed the compliance burden for AI ad labels on advertisers rather than on AdSense publishers. Today's notice keeps that structure and widens the map.

The buying side moved first again. According to Search Engine Roundtable, Google updated its Google Ads help page on AI content labels by November 5, 2026 to add California, South Korea and Vietnam to the locations where ads with assets designated as AI created or edited also carry a visible overlay. That page now lists six locations: the European Union, India, New York, California, South Korea and Vietnam. It ties the overlay to campaigns that target those locations.

Six regions on one page, three on another

Google's own documentation is not yet consistent. The Display & Video 360 help page on AI content label settings, checked today, still lists only the European Union, India and New York as locations where labels appear on the ad itself. The AdSense notice names Display & Video 360 as one of the two sources of labeled creatives in the three new regions.

There is a second gap. The Google Ads documentation keys the overlay to the locations a campaign targets. The AdSense notice speaks of labels appearing "in these regions." Neither document explains how a campaign's location targeting maps onto an impression on a publisher's page, for example when a publisher based outside California serves an ad to a user inside the state. Google has not published that detail.

The laws behind the list

Google describes the requirements in all three places as "upcoming," without saying which specific rules prompted the expansion. Two of the three jurisdictions already have AI labeling laws in force, although further implementing rules in each may still be pending.

California

California's most direct rule for advertising is SB 1050, which Governor Gavin Newsom signed on September 16, 2026. According to an October 15, 2026 analysis by the law firm Mayer Brown, its obligations take effect on January 1, 2027. The law requires a clear and conspicuous disclosure when an advertisement prominently features a synthetic performer, defined as a digital figure, voice or representation created wholly or partly with generative AI, with wording substantially similar to a statement that the performance features a synthetic performer. Unlike New York's law, it covers audio as well as visual and audiovisual ads, according to Mayer Brown.

The California law matters to publishers in one specific respect. Advertising mediums, including broadcasters, platforms, streaming services and ad networks, must take down a violating ad only after a court has found that it breaks the law and the order reasonably identifies the ad, according to Mayer Brown. The law carries no civil penalties of its own; a violation can be treated as false advertising, enforceable under California's Unfair Competition Law.

California follows New York. That state's synthetic performer disclosure rule, an amendment to section 396-b of the General Business Law, took effect on June 9, 2026 with penalties of $1,000 for a first violation and $5,000 for subsequent ones. Newspapers, broadcasters, streaming services and other media carriers are not liable for publishing a non-compliant ad under the New York law.

South Korea

South Korea's Framework Act on artificial intelligence, commonly called the AI Basic Act, took effect on January 22, 2026, according to the Future of Privacy Forum. Article 31 requires providers of generative AI to indicate that output was generated by AI, and to notify or display that fact clearly when synthetic sound, images or video are difficult to distinguish from reality. Fines under Article 43 reach 30 million won.

Advertising-specific measures have followed. In December 2025, the government said it would require advertisers to label ads made with AI and would make the companies operating platforms responsible for ensuring that advertisers comply, according to the South China Morning Post; the plan involved revising the country's telecommunications law. Separately, revised Korea Fair Trade Commission guidelines on endorsements took effect on June 1, 2026, requiring marketers to state prominently when an endorser is a virtual character and, in photo and video ads, to keep the label visible near the AI figure while it is on screen, according to Inside Retail Asia.

Vietnam

Vietnam's Law on Artificial Intelligence, No. 134/2025/QH15, was adopted on December 10, 2025 and took effect on March 1, 2026, according to the text on the government portal that Google's notice links to. Article 11 requires AI-generated audio, images and video to be marked in a machine-readable format by providers. Deployers must notify the public when AI-generated or edited content is capable of causing confusion about its authenticity. AI-made audio, images or video that simulate a real person's appearance or voice, or recreate real events, must carry a clearly recognisable label. The government is to set the forms that notification and labeling must take. The labeling provisions apply to AI content generally rather than to advertising in particular.

What the documents leave open

Several questions remain unanswered by the two pages.

The start date is a month, not a day. The phrase "starting November 2026" leaves publishers in the three regions without a fixed point at which labeled creatives begin to appear.

What will the label actually look like? Neither page shows the overlay on a display unit, a video ad or an app placement, or says how much of the creative it covers. For advertisers, Google's July documentation recommended keeping custom labels outside a margin of about 5.5 percent around an asset's perimeter. The AdSense pages say nothing about where Google's own overlay sits.

Blocking is not addressed. The guidance does not say whether publishers can filter or block creatives that carry AI labels, or whether labeled ads are treated any differently in the auction.

The native route raises a placement question. European rules on AI disclosure set a high bar for visibility. The European Commission's materials call for labels to be visible at the latest at first exposure, and practitioner guidance reported by PPC Land when the Commission published its free AI labelling icons held that a footer disclosure does not meet that standard. Google's guidance for third-party native ads allows the AI disclosure to sit on a page reached through a privacy link. Whether such a disclosure satisfies a rule that demands visibility at first exposure is a question the help page does not take up.

Why it matters for publishers

For most publishers, the change will be visible rather than operational. Labels on Google-bought ads are applied upstream, and AdSense publishers have no setting to manage them. The harder questions sit in the other two rows of Google's table.

Legal guidance in Europe has so far drawn a line that favours publishers. Under the EU AI Act, whose Article 50 transparency obligations have applied since August 2, 2026, the labeling duty sits with deployers, and the Commission's guidelines published on July 20 named advertisers, agencies and media outlets among them, with fines of up to EUR 15 million or 3 percent of worldwide turnover. A September 2026 guide from IAB Austria concluded that a publisher which passes an ad through unchanged generally carries no labeling duty, but becomes a deployer if it alters or personalises the ad with AI. Dutch trade body VIA Nederland reached a similar view of the agency's role, mapping four scenarios in which disclosure is triggered.

Google's table follows the same logic: whoever builds and buys the creative carries the label. Yet the format guidance shows where the technical burden still lands. A video publisher on PAL must build icon rendering into its player for any third-party label to appear at all. A custom-badging publisher must turn a flag in the ad response into a visible marker. A publisher selling direct must confirm that every creative it traffics carries a disclosure where one is needed. None of that is automatic.

The geography is also widening faster than the documentation. In 2025, China made AI content labeling mandatory from September 1, with explicit labels required on text, images, audio, video and virtual scenes. In July 2026, Google signed the EU Code of Practice on Transparency of AI-Generated Content. Now three more jurisdictions join the overlay list, each with a different definition of what must be labeled: synthetic performers in California, generative AI output and virtual endorsers in South Korea, and content that could cause confusion about authenticity in Vietnam. Google's labels follow its own list. Publishers' obligations follow the law.

Timeline

Summary

Who: Google, through its publisher products AdSense, Ad Manager and AdMob, and the publishers that use them. Advertisers buying through Google Ads and Display & Video 360, third-party buyers using Authorized Buyers, Open Bidding and SDK Bidding, and creative vendors are also affected.

What: Google is extending visible AI labels to ads served on publisher properties in California, South Korea and Vietnam when the ads come from Google Ads or Display & Video 360. A companion help page says Google publisher products do not label third-party creatives automatically, and sets out display, video and native methods for adding disclosures, including the VAST Icons element and an OpenRTB privacy link.

When: The notice was posted today, November 12, 2026, and says the labels start in November 2026 without naming a day. The Google Ads help page added the three regions by November 5, 2026.

Where: California, South Korea and Vietnam, joining the European Union, India and New York, where on-ad labels already applied. The notice was published on the AdSense Help Center.

Why: Google cites upcoming regulatory requirements in the three jurisdictions without naming them. California's SB 1050 on synthetic performers in ads takes effect on January 1, 2027, South Korea's AI Basic Act has applied since January 22, 2026, and Vietnam's Law on Artificial Intelligence since March 1, 2026.