AI is outrunning who controls it

The IAB Europe 2026 study of 50 industry respondents found 86% already use AI for marketing, but the gap between optimization and autonomy is wide: 59% use it for programmatic optimization and 55% for media planning, while agentic buying and selling sits at just 38%. Expectations run ahead of deployment, with only 30% expecting agents to become a main buying method in some markets next year and no forecast commanding a majority. Human oversight is thin: of 47 respondents, 36 report either no agentic system in daily use or one at the two lowest autonomy levels. Digiday's Ronan Shields frames the same tension through the supply side, where a MonetizeMore LinkedIn post reignited debate over Google Ad Manager's newly documented option to set bidder-specific price floors, framed as the return of a lever publishers lost when unified pricing rules arrived in 2019. The survey's verdict is blunt: respondents rated AI in current ad operations just 2.76 out of five, with 69% citing operational efficiency and only 3% citing fill rate or yield. As Viant CEO Tim Vanderhok put it on the agentic-commerce front, "Meta has spent years learning what its users are interested in and how to create demand. Now, by taking aim at the checkout experience, Meta is attempting to collapse the entire funnel into one interface."

The crux

Programmatic is being automated faster than its governance, and the power to inspect and override an AI buying decision is not being built alongside the automation, which means whoever controls the agent's guardrails controls the auction.

Sell-side read

Publishers should watch bidder-specific floor pricing closely: if the Google remedy final judgment keeps per-bidder floors, it is a genuine yield lever against bid shading, but only for publishers who can measure net revenue rather than gross bid.

Watch

Whether the final Google ad tech remedies judgment text preserves per-bidder floors, and whether AAMP's pricing provenance field gets adopted before agents hit scale.