Skip to content

Insights

What Every Advertiser Should Know About AI-Driven Media Platforms

Gabriella Buzzelli, Vice President, Performance MediaPublished Updated 4 min read

Google, Meta and Microsoft are automating more of your media with AI. Where the brand-safety and performance risks sit, and how to keep control.

A grid of ad tiles moved by automated gears and data streams, protected by a translucent shield

The media platforms driving digital advertising today (Google, Meta, Microsoft and others) are changing faster than at any point in the last decade, and most of those changes are being driven by AI. That isn’t inherently a problem, but it does call for a more deliberate conversation about how and when these tools belong in your media mix.

Key takeaways

  • AI campaign types like Performance Max and AI Max decide placements and creative for you, using signals you can’t fully see.
  • Platform terms make you responsible for every auto-generated asset that runs.
  • Before leaning in, answer two questions: how much brand-safety risk can you accept, and how much early performance volatility can your business absorb?

Why this is happening

The major platforms have made enormous bets on AI infrastructure, and they need to show returns. Google’s search advertising business, the most reliable revenue engine in digital media history, is approaching a ceiling. There’s only so much growth available when every advertiser is competing in the same auction. Products like Performance Max, AI Max and Microsoft’s AI-driven campaign types represent a new growth surface for these platforms and a way to diversify their revenue. They’ve proven genuinely useful in the right contexts, but they’re being rolled out aggressively, and that pace creates gaps.

What’s actually changing

These products work differently from traditional campaign structures. Rather than the advertiser defining precise targeting, placements and creative, the platform takes inputs and makes decisions autonomously, optimizing toward an outcome using signals the advertiser can’t fully see, drawn from thousands of data points across device, identity and auction-based touchpoints. In practice, assets can be auto-generated and dynamically combined, placements can expand beyond what was manually selected, and settings can shift with platform updates that aren’t always clearly communicated.

Industry reporting backs this up. Microsoft Advertising autogenerates text assets in responsive search ads, adding copy the advertiser never wrote. AdBacklog noted that the Microsoft Audience Network (MSAN) is automatically enabled in search campaigns with no master toggle to opt out, and that advertisers often discover changes after delivery has already shifted. Academic research on AI advertising platforms has made a similar case against their opacity, arguing advertisers need clearer explanations of how the systems make decisions. Beyond the formal reporting, dozens of threads across industry forums and social media describe the same issues, and most of them never make it into a published article.

Where the real risk lives

Not every advertiser faces the same exposure, and how you approach these platforms should reflect your situation. Two questions are worth thinking through carefully.

The first is brand safety and asset control. How much does it matter that every impression, placement and creative execution meets your brand standards? For some categories this is a preference. For others, such as pharmaceuticals, financial services, other highly regulated industries and precision B2B brands, it’s a hard requirement.

It’s also worth understanding where liability sits when automation produces something that shouldn’t have run. Platform terms and conditions are written to protect the platform. Google’s updated Terms of Service, announced in April and taking effect July 1, make this explicit, stating that advertisers have a “continued obligation to review, approve, or remove all campaigns and ad assets generated automatically by Google Ads features.” In other words, the platform reserves the right to generate assets on your behalf, and you are responsible for everything that runs. If an auto-generated asset conflicts with regulatory requirements, misrepresents your product or simply doesn’t meet brand standards, the recourse available to advertisers is limited. We can request removal and put exclusions in place going forward, but recovering costs or seeking compensation for what already ran is not a realistic expectation. For brands in regulated categories especially, that isn’t a theoretical risk. It should factor directly into whether and how these tools belong in the mix.

The second is performance uncertainty. These platforms optimize over time, and the learning period has proven necessary. Results can be inconsistent early before they level out, and, frustratingly, the optimization logic isn’t always transparent. How much variability can your business absorb while the platform finds its footing?

We work with our clients to answer those two questions, and the answers should drive how aggressively we engage with these tools.

What to do about it

Every campaign running on these platforms carries some exposure to the issues described above. How much depends on how campaigns are set up, what’s been excluded and how actively they’re monitored. The good news is that with the right setup discipline and ongoing oversight, the performance upside these tools offer is accessible without the brand risk.

If you’re running media on any of these platforms and haven’t had a candid conversation about where your campaigns are exposed, now is the time. MODCO works with brands across categories to audit current setups, identify gaps and build the guardrails that let you capture the performance upside without the brand risk. Talk to us to get started.

A search bar surrounded by floating answer cards, with a cursor that never reaches a link

SEO Didn’t Die. The Click Did.

Most Google searches now end without a click. What 18 months of AI search changed for SEO, what it didn’t, and how we measure visibility heading into 2027.

Ali McDonaldUpdated

Now, let's connect and continue the journey together

Contact