Google has announced a new feature that will inform users when an advertisement has been created or edited using generative artificial intelligence. The label, accessible through the My Ad Center panel, will appear across Google Search, YouTube, and Google Discover, and is being deployed globally. The move is part of a broader push for transparency in digital advertising, as AI tools become cheaper and more accessible, making it easier to produce synthetic product imagery that could mislead consumers.
The disclosure can be found by clicking the three-dot menu or the info icon on any ad, which opens the My Ad Center panel. Previously, this panel allowed users to block or report ads and to understand why a particular ad was shown to them. Now it adds an option labeled “how this ad was made,” which surfaces any AI involvement in the ad's creation or editing process. The rationale behind the feature is straightforward: AI enables the generation of high-quality, realistic product images at minimal cost, which can deceive shoppers into believing they are looking at actual photographs rather than synthetic representations.
The Honor-System Catch
While the feature is a step toward greater transparency, its effectiveness depends heavily on how the ad was produced. When advertisers use Google’s own generative AI ad tools, the disclosure is switched on automatically. However, when an ad is created using third-party tools or platforms, the advertiser must actively flag that AI was used. Google has stated that it will not independently verify these claims, meaning the label relies entirely on the honesty of advertisers. This reliance creates a significant gap: an advertiser hoping a synthetic scene passes for a genuine photo has little incentive to voluntarily disclose its AI origins.
The same incentive problem plagues other areas of online content. For example, social media platforms have struggled with misinformation for years, and studies show that users often fail to notice even clearly labeled synthetic media. Google’s system, which is entirely voluntary, may face similar challenges. The company has not indicated any plans to audit advertisers for non-compliance, nor has it specified penalties for failing to disclose AI involvement. This contrasts with its approach on YouTube, where the platform will auto-label AI-generated videos regardless of whether creators disclose them, using technical detection methods.
Regulatory Pressure and the EU AI Act
The timing of Google’s announcement is not coincidental. The European Union’s AI Act, which includes strict transparency obligations for AI-generated content, is set to begin enforcement in August 2024. Under the Act, providers of AI systems that generate or manipulate image, audio, or video content must clearly label outputs as artificially generated or manipulated. This applies to a wide range of applications, including advertising. Industry groups, particularly retailers and advertisers, have been lobbying to exempt AI-made ads from these rules, arguing that mandatory disclosure could stifle creativity and impose unnecessary costs.
By introducing a voluntary, self-declared labeling system, Google is positioning itself as proactive, potentially softening the need for more aggressive regulation. The move is part of a broader industry trend where tech companies preemptively adopt transparency measures to shape the regulatory landscape. For instance, Meta (Facebook) and TikTok have rolled out similar labels for AI-generated political ads. However, Google’s extension to all commercial ads goes further than most of its peers, covering not just political but product and service advertisements.
The EU AI Act is expected to influence global standards, as companies operating in Europe must comply regardless of where they are headquartered. Google’s global rollout of the AI ad label means it will likely become the default standard even in markets without such requirements. This could create a competitive pressure on other ad platforms, like Amazon or Microsoft, to adopt similar disclosures.
Technical and Operational Details
From a technical perspective, the new label is integrated into Google’s existing ad ecosystem. Advertisers using Google’s own generative tools—such as those in Google Ads that create images or copy for product listings—will automatically have the disclosure applied. The detection is based on metadata embedded during the ad creation process. For ads created externally, advertisers must check a box in the ad creation interface indicating that AI was used. Google has not specified how it will enforce this, nor how it will handle cases where an advertiser claims to have used AI but does not disclose it.
The labeling is visible to users on desktop and mobile, though its placement and visibility may vary. Google has not provided examples of what the label will look like, but it is expected to be a small text or icon near the ad. Users can also find a full explanation in the My Ad Center panel, which includes a history of ads they have interacted with and options to customize ad preferences.
The feature builds on Google’s earlier requirement for AI disclosure on election ads. That policy was implemented in late 2023, just ahead of major elections in India, the United States, and the European Union. The expansion to commercial ads reflects a growing awareness that AI-generated product imagery can be just as deceptive as political advertising. For example, a fake photograph of a product that does not exist, or a heavily edited video of a product in use, could lead to consumer fraud or disappointment.
Industry Reactions and Concerns
Reactions to the announcement have been mixed. Consumer advocacy groups have praised the move as a necessary step, but they caution against the reliance on self-disclosure. Without enforcement, it’s just a placebo, said one analyst quoted in trade press. Advertisers, meanwhile, have expressed concerns about the potential for confusion among consumers who may not understand what AI involvement means. Some worry that the label could stigmatize AI-assisted advertising, even when the technology is used for benign purposes like correcting lighting or removing blemishes.
Small businesses that rely on AI tools to create professional-looking ads may also face a dilemma: if they disclose AI use, they risk appearing less trustworthy, but if they don’t, they risk violating Google’s new policy—though without enforcement, the risk is minimal. The lack of verification also means that bad actors can simply ignore the requirement, reducing the label’s effectiveness.
There is also a technical challenge: as AI detection methods improve, so do methods to evade detection. Google’s own AI systems could theoretically be used to check ad imagery for signs of generation, but the company has not made that commitment. In contrast, on YouTube, Google uses automated systems to detect AI-generated content, but has acknowledged that detection is not foolproof. For ads, the company appears to be taking a less invasive approach, perhaps to avoid alienating advertisers.
Broader Implications for Synthetic Media
The debate over labeling AI-generated content extends far beyond advertising. News articles, social media posts, music, and even scientific images are increasingly generated or manipulated by AI. Governments and international bodies are scrambling to create frameworks for authenticity. The EU AI Act, the U.S. Executive Order on AI, and China’s regulations all mandate some form of labeling. However, definitions vary: what constitutes “AI use” can range from full generation to minor editing, making compliance complex.
Google’s label applies only to ads that have been “created or edited with generative tools.” This excludes content that was simply optimized by AI algorithms, such as automated bidding or targeting. The distinction is important because many advertisers use AI for back-end processes without altering visible content. The label specifically targets content that could deceive a viewer.
Another issue is consistency across platforms. While Google, Meta, and TikTok have implemented labeling for AI-generated political ads, their approaches differ. Google’s voluntary commercial ad label is the first of its kind on a major ad network. If it proves successful, it could become an industry standard. However, if the honor system is widely abused, it could backfire and erode trust in all such labels.
The technology for generating realistic images has advanced rapidly in the past year. Tools like Stable Diffusion, Midjourney, and DALL-E can create photorealistic scenes that are indistinguishable from authentic photos to the untrained eye. This raises the stakes for ad transparency. A consumer might look at a furniture ad showing a room with perfect lighting and assume it is a real photograph of an available product, but the room could be entirely synthetic. The product itself might also be generated, leading to differences in quality or appearance upon delivery.
Google’s decision to extend the label to commercial ads is a recognition that such deception is increasingly possible and profitable. The company profits from ad sales, so there is an inherent conflict of interest: too much transparency might deter advertisers. Balancing these forces is delicate. The voluntary nature of the label may be a compromise—enough to satisfy regulators while not burdening advertisers.
The rollout will be monitored closely by EU officials, who have the power to mandate stricter rules if industry self-regulation fails. The AI Act includes provisions for fines of up to 7% of global annual turnover for violations, so the incentive for Google to appear compliant is strong. Other regions, including the UK and Japan, are also considering similar laws, meaning the landscape for AI labeling is likely to become more uniform in the coming years.
For now, the success of Google’s feature will depend on how many advertisers choose to disclose. Early adopters may include brands that want to position themselves as transparent. Conversely, those who rely on deception will have no reason to participate. The result could be a two-tier market: honest advertisers get labeled, while dishonest ones remain indistinguishable from traditional ads. This is exactly the problem the label was designed to solve, yet the solution is only partial without enforcement.
In the absence of verification, users must remain skeptical. Google has provided a tool to ask how an ad was made, but the answer is only as reliable as the one who provides it. As the ad industry navigates the age of generative AI, the balance between innovation and trust will require more than a checkbox. It will require robust systems to ensure that the label means what it says.