difrnt. Radar: AI Is Everywhere. Control Is Nowhere.

Some weeks, a single story dominates the marketing headlines. Other weeks, everything moves at once, in different directions, but with a common thread. This is the second kind.

AI is no longer just a conference talking point. It's operational. It's in the media buying systems managing billions. It's in the tools generating TikTok ads. It's in e-commerce checkout. And, perhaps most significantly, it's in the behavior of the consumers buying our products.

At every step of the funnel, the same question: who's making the decisions? The AI, the platform, or the marketer? Here are four developments from the past few days that, taken together, paint a clear picture. AI has occupied every stage. Strategic control remains yours, but only if you actively exercise it.

70% of media budgets will flow through AI. Not for your benefit.

A Gartner report cited by Marketing Dive puts numbers on what many of us already sensed: by 2028, over 70% of global ad spending and 80% of U.S. ad spending will flow through AI-powered self-serve platforms. Google, Meta, and Amazon already control more than half of American media buying and are approaching 60% globally.

What does this mean in practice? The platform algorithms don't optimize for your business objectives. They optimize for their own quarterly revenue. When a platform recommends "You should spend more with us," it's not consulting. It's selling.

We see this constantly with clients: platform-reported performance looks great, but CRM or Google Analytics 4 numbers tell a different story. Platforms carry a structural bias, not because they're malicious, but because they're built to maximize usage of their own ecosystem.

The fix isn't to abandon automation. It's to stop leaving it unsupervised. Independent measurement through third-party providers, a human who asks "why?" before approving each recommendation, and a clear separation between platform reporting and business reality. Otherwise, you're not doing media buying. You're doing media donating.

This is the same trend we flagged about Meta: platforms are taking over control, and marketers need to consciously decide what they delegate and what they don't.

TikTok says anyone can run AI ads. How true is that?

Social Media Examiner documents how TikTok launched its Agentic Hub and MCP protocol, allowing marketers to connect AI tools directly to Ads Manager. No code, plain language: "Show me last week's campaign performance" or "Create a new ad set for the 25-34 audience."

Beyond campaign management, Symphony Creative Studio promises over 100 videos per week generated from product images. The number sounds impressive, but the details temper the enthusiasm: AI gets the creative to roughly 70% of where it needs to be. The remaining 30% requires human refinement. And AI-generated presenters still lack credibility, especially for brands that have built their identity on authenticity.

What actually works on TikTok? Content that doesn't look like an ad. Videos where someone does something naturally (applies a product, demonstrates, tells a story) outperform those filmed straight to camera with a sales pitch tone. AI helps with production and scaling creatives, but it doesn't replace content instinct and audience knowledge.

The implication for agencies and brands: TikTok's AI tools lower the barrier to entry, and that's a good thing. But "easier to make" doesn't mean "easier to make well." Without a content strategy, 100 videos per week means 100 misses per week. With strategy, AI becomes a real multiplier.

Your product is in ChatGPT. But can your checkout keep up?

Search Engine Journal raises a problem few are discussing: it's relatively easy to make a product visible in AI search results. The hard part comes after.

AI shopping orders have surged 15x on Shopify since early 2025. Three new protocols (Google Universal Commerce, OpenAI Agentic Commerce, Salesforce Agentforce) connect AI agents directly to checkout. But the way these agents buy is fundamentally different from how humans browse. They make rapid, parallel API calls in fractions of a second. And that breaks systems built for slow, linear human sessions.

The core problem is deceptively simple: inconsistent data. The price in the AI catalog doesn't match the checkout price. Inventory shows available in one system and unavailable in another. Refunds don't process correctly. Three tests every retailer should run now: load-test checkout APIs with rapid parallel calls, verify product data accuracy as machines read it, and validate refunds end-to-end.

We've written before about how AI might know nothing about your brand. But even when it does know you and recommends you, a checkout that can't keep up with AI speed turns an opportunity into a lost customer.

Consumers using AI spend nearly double

An NRF and Prosper Insights study cited by Marketing Dive reveals a fascinating paradox from back-to-school season: parents using AI and other digital tools for shopping plan to spend $737 per child. Those using no technology: $381. Nearly double.

46% of parents are using AI to compare prices, and 53% use it specifically to find the best deals. Meanwhile, loyalty program usage jumped 7% year-over-year to 46%, a sign that price sensitivity remains high even among those spending more.

But perhaps the most interesting insight: 91% of parents with teenagers say their children influence their brand preferences. Gen Z isn't just consuming. They're actively influencing family budgets.

For marketers, the implication is direct: the consumer who arrives through AI isn't an ordinary consumer. They're more informed, compare more, have higher expectations, and paradoxically, spend more. Not because they're impulsive, but because AI gives them confidence in their purchase decision. Anyone still targeting purely on traditional demographics is missing a more relevant segmentation: the buyer's digital behavior.

Four stories from the past week, one thread: AI is no longer an optional feature you toggle on when convenient. It's already in media budgets, in creative production, in checkout, and in customer behavior. Every stage of the funnel now has an AI component, visible or not. The question is no longer "Are we using AI?" but "Do we control how AI uses us?" And, more importantly, "Do we have the people to exercise that control every day?"