Google Ads will use generative AI to resize Performance Max video ads

Google Ads is rolling out generative AI that automatically resizes and reformats Performance Max video ads, so a single upload can fill far more placements across YouTube, the Google Display Network, and beyond. Search Engine Land reports the change, and the detail that matters most for advertisers is the rollout mechanics: the feature is on by default, with a September 4 opt-out deadline for anyone who does not want Google reshaping their creative.

What Google is changing

Performance Max campaigns already spread a single set of assets across every Google surface, using automation to decide where and how they show. The new layer lets Google's models take one video and generate alternate aspect ratios and lengths, cropping, restitching, and reframing footage to match vertical, square, and horizontal slots. Dataconomy describes it as generative AI optimization applied directly to video, so ads built for one format can be spun into the shapes each placement demands without the advertiser producing them.

Why it is on by default

The default-on setup is the part media buyers are flagging. WebProNews notes that Google is turning generative AI loose on video creative automatically, which means brands that take no action will have machine-edited versions of their spots running by early September. Opting out requires going into campaign settings before the September 4 cutoff. Teams that manage many accounts will need to audit each one rather than assume the status quo holds.

The creative control and brand safety questions

Automatic reframing can lift reach, but it hands editing decisions to a model. A crop can push a logo, a caption, or a legal disclaimer out of frame, and a shortened cut can drop the line that made the ad work. For regulated categories the stakes are higher, since a reframed asset might no longer meet the placement rules it was cleared under. Brand teams that spent months on a hero video may find AI variants live before anyone signs off on them.

For marketers the practical move is simple: decide now whether to keep the feature or opt out, then review the generated variants the same way you would review any new cut. This fits the wider pattern of platforms automating creative work, a shift we cover in our guide to AI for marketing. It also lands in a busy stretch for Google video, following its recent change to how YouTube counts views, and against a backdrop of rising scrutiny of how platforms use engagement mechanics, seen in the Meta youth addiction trial. For a broader view of putting AI to work well, Techpresso's roundup on AI for productivity is a useful companion.