AI-generated ads: Who is in control?

Who is actually writing your ad copy? A fashion retailer is preparing its Black Week campaign. The headline, discount, and mandatory disclosures have all been coordinated and approved. Yet, the ad that actually goes live looks different: the advertising platform has automatically created a new headline and selected a different landing page. There was no separate approval for these variations.

What sounds like a loss of control is now part of the technical capabilities of AI-powered advertising platforms. Google, Meta, TikTok, and ChatGPT are increasingly automating the creation, adjustment, and selection of ad assets. This opens up opportunities for more efficient campaigns, but it also changes the requirements for brand management. Who controls the content, which statements are allowed to be automatically altered, and where are binding boundaries needed?

More automation does not automatically mean better results. The difference lies in how brands use which functions and how targeted they are. We show what possibilities the platforms offer, what risks arise, and how brands can keep their advertising under control.

Key Takeaways

• AI is changing ad creation: Google AI Max is enabled by default for new search campaigns. OpenAI, Meta, and TikTok also offer features that automatically create, adjust, or select ad copy and creatives.
• Brand communication is becoming more demanding: Automatically generated content can deviate from approved messaging. Outdated product information, inappropriate phrasing, or changed landing pages pose risks.
• Control remains possible, but not seamless: Advertisers can specifically manage or deactivate many AI functions. Which adjustments and actually displayed variations they can review differs depending on the platform.
• Automation requires clear guardrails: Companies should monitor AI-generated ad copy, especially for sensitive content. Reliable source data, binding guidelines, and regular audits are crucial.

When the platform becomes the copywriter

AI-generated ads are long past being just a choice between pre-created variations. Platforms formulate additional text, adapt content, and combine ad assets automatically. The extent to which they do this varies significantly.

Google: AI Max is an optimization layer for search campaigns and is enabled by default for new campaigns. With text adjustment, Google can create additional headlines and descriptions based on existing ads, keywords, and landing page content. Which functions remain active can be controlled in the campaign settings. With Final URL Expansion, Google can also select a different landing page. If Google selects an alternative URL, pinned assets may not be taken into account. This can be controlled via text guidelines, keyword exclusions, and URL exclusions, among other things. With AI Brief, Google is also testing a briefing in its own words, which is now also available in German in the closed beta.

OpenAI: ChatGPT Ads offers AI-powered ad asset optimization that can adapt headlines and descriptions to the conversation context and translate ad copy. According to current product documentation, adjustments to image sizes and destination URLs are also possible. OpenAI has announced that these features will be enabled by default starting October 31, 2026. Advertisers can continue to deactivate them at the account or campaign level.

Meta: Advantage+ Creative can automatically create and adjust ad variations. Branding guidelines and excluded terms are intended to help manage brand communication. Which functions are available depends on the respective setup.

TikTok: Smart+ automates, among other things, the selection and combination of creatives. TikTok has expanded the preview functions for automated ads. Which combinations can actually be checked before the campaign starts depends on the available features and the respective setup.

What does this mean for the brand voice?

With automation, a portion of control shifts: away from the finished ad and toward the information from which it is created.
Google, for example, uses content from landing pages to create new ad copy. If those pages contain outdated product information or expired promotional offers, this can lead to problematic advertising claims. When many brands use similar optimization mechanisms, their ads can become linguistically interchangeable. What is optimized for clicks in the short term does not automatically strengthen a brand's recognizability.

This poses a new question for marketing managers: How much creative freedom should we give the platform without losing our brand identity?

Who bears the responsibility?

Beyond brand impact, the responsibility for false or misleading AI-generated statements is also coming into focus. A ruling by the Higher Regional Court of Hamm highlights the legal questions that arise in this context.

The following assessment does not constitute legal advice.

On May 12, 2026, the Higher Regional Court of Hamm ruled (case no. 4 UKl 3/25) that a company must be held accountable for misleading statements made by its own AI chatbot. The chatbot had provided false information regarding medical qualifications. The ruling concerns a company-owned chatbot. Whether the same principles apply to ad copy generated by external advertising platforms has not yet been definitively settled.

Regardless of the still-unresolved legal classification of external platform ads, advertisers should carefully monitor automated ad materials, particularly regarding pricing, health claims, sustainability assertions, and warranty terms.

Five guardrails for automated ads

Determining the right level of automation is a daily task in campaign management. For Laurids, the focus is on balancing performance with brand control:

"Automation is useful where it makes processes more efficient and improves campaign performance. As soon as it comes to brand messaging and the quality of delivery—such as which target group sees which copy—human oversight remains essential. The general rule is: maintain as much control as possible and use as much automation as adds value to the campaign."

At the same time, automation can offer concrete performance advantages, as Laurids explains:

"Many ad formats, for example in Google Ads, benefit in terms of ad rank when automations are enabled. This has a direct impact on cost-per-click. While the ads then require more intensive management, they also deliver good results."

This leads to five guardrails that we follow in campaign management:

1. Know your active automations. Before launching a campaign, check which AI features are enabled, what content they can modify, and which settings can be controlled individually.

2. Decide based on risk and category. Lifestyle campaigns can handle more leeway than promotions with binding price information or advertising for regulated products. The more sensitive the claim, the stricter the requirements.

3. Set boundaries. Keyword exclusions, tone-of-voice guidelines, and restrictions on landing page selection limit undesirable statements. Binding formulations require additional oversight.

4. Keep source data up to date. When platforms use website and product information to generate new copy, outdated content becomes an advertising risk. Landing pages and campaign messaging must be aligned.

5. Monitor actual ad delivery. Clicks and conversions aren't enough. Regularly check generated assets, search terms, landing pages, and the available ad variants. The level of insight you get varies by platform.

The transparency problem remains

Not every platform fully discloses which ad variant was served in which context. Google offers reports on search terms, assets, and landing pages, and has announced a unified view. Regarding ChatGPT Ads, Dentsu reported limited reporting insights in September 2026: performance metrics were available, but the specific conversational context of individual ad deliveries was not. TikTok, by contrast, focuses more on previews before a campaign launches.

At adlicious, monitoring newly generated assets is therefore part of daily campaign management. Laurids and his team check not only performance but also the resulting ad variants and their impact.

"Daily monitoring also helps us make ad copy optimization more efficient and use AI as a partner for writing and optimization," explains Laurids.

Conclusion

The fact that brands are using AI-generated ads is hardly up for debate anymore. The more interesting question is under what conditions this makes sense. The more sensitive the advertising message, the stricter the guidelines and control processes must be. Those who understand the possibilities and limitations of individual platforms can use automation in a targeted way without relinquishing control over their brand communication.

How much control do you have over your AI-generated ads?

Which automations are active in your campaigns? Which messages can the platform change, and where are binding guidelines required?

Together, we will review your campaign setup, identify potential risks, and show you how to meaningfully combine automation with your brand requirements.

Your contact person is Florian Meyer, Head of Sales at adlicious.

Schedule an appointment with Florian →

Read more

• ChatGPT Ads: When ChatGPT becomes an advertising space.
• How dynamic ads generate thousands of variants without the brand losing control over its messaging.
• Why short-term performance and brand building belong together and how brands can remain measurable in the process.
• How data-driven dynamic ads increase cinema attendance.