When the platform chooses the asset, creative approval becomes part of media buying.
Creative is no longer the thing you approve before the campaign starts.
TikTok says its Auto-Selection feature for Lead Generation can source, test, and scale top-performing assets from brand uploads, creator content, and AI-generated content. Its catalog auto-crawl can also find eligible product images and videos on public landing pages. Meta is testing brand-aware generation, a stored brand memory, and a built-in approval workflow for creative modifications. These are two independent platform announcements, but they point to the same operating change: the creative pool can expand after the team thinks the brief is closed. TikTok's Q3 product preview and Meta's Cannes announcement make that shift explicit.
The decision is not whether to use automation.
The decision is whether your approved creative system can survive it.
The Old Approval Happened Too Late
The incumbent process treats creative review as a finish line.
A team writes the brief. Design builds a set of assets. Brand approves the copy. Legal clears the claim. Media launches the campaign. The governance file closes because the work is live.
That process fit a world where the launch set was the delivery set.
It breaks when the platform can choose from creator posts, pull product images from a landing page, make a new variant, or reintroduce an asset that performed well in a different context. The team may still have an approval step. It no longer controls the full set of things the system can put in front of a buyer.
A polished review process can still leave an uncontrolled media pool.
More Inputs Create More Ways To Be Wrong
The first cost is claim drift. A product page can hold an old price, a broad capability statement, or a photo that no longer matches the offer. If that page becomes an automated source, stale site content becomes paid creative.
The second cost is context loss. A creator clip can be appropriate in one audience, geography, or product category and misleading in another. A predicted-performance score does not know the commercial boundary your team intended.
The third cost is reporting confusion. If the platform adds, excludes, and tests assets over time, the winning campaign is not proof that the original brief won. It is proof that a changing pool produced a result. That distinction matters when a marketing director decides what to fund again.
Get one wrong and a weak claim reaches more people. Get two wrong and the team cannot explain which asset or rule changed the result. Get three wrong and a brand review becomes theater while the media system runs somewhere else.
That is not a creative problem. That is a control problem.
Turn The Brief Into A Control Layer
The answer is not to ban every automated feature. Automation can find useful combinations faster than a weekly status meeting. The answer is to make the approved brief usable by the systems that select and generate creative.
Start with four controls that live before campaign setup:
- A claim library that states which promises are approved, what evidence supports each claim, where it may appear, and what version expires.
- An asset register that identifies each approved image, video, creator asset, product page, audience context, geography, and offer it can support.
- An exclusion rule that lists what the platform must never select, crawl, regenerate, or place into a test without a named human review.
- A change log that records when the asset pool changes, who approved the change, and which reporting window began after it.
Then assign ownership. Brand owns the claim and the expression. Product or operations owns price, availability, and product accuracy. Legal owns regulated boundaries. Media owns delivery settings and exclusions. Analytics owns the test record and the outcome definition.
One team cannot do all of that alone. One source of truth can make the handoffs visible.
TikTok's announcement is useful because it describes both the expansion and the controls. Advertisers can add or exclude assets through Custom Selection, and can review predicted assets before excluding them. Meta's announcement is useful for the same reason. It frames brand memory and approval workflow as part of the creative system, not as a file someone reads once. Those features remain platform-specific and, in Meta's case, in testing. The operating lesson is broader than either product.
The more adaptive the delivery system becomes, the more explicit the input standard must become.
Control Or Theater
Marketing leaders now have two paths.
One path treats automation as a production shortcut. Feed the platform every asset available, approve a few examples, then judge success from the campaign dashboard. That is not creative governance. That is broad permission with better reporting.
The other path treats the creative pool as a media setting. It defines the claims, assets, exclusions, owners, and evidence that can enter the system. It logs changes, measures outcomes after those changes, and removes expired material before an algorithm finds it again.
That is not slower media buying. That is trustworthy media buying.
Magnet connects brand, website, paid media, analytics, and operations into one demand system. We turn approved strategy into controls the work can actually follow.
Build a demand system with Magnet.
Sources
- TikTok For Business, TikTok Product Preview: What's New For Brands In Q3 2026
- Meta for Business, Cannes Lions 2026: Meta's New AI Creative & Creator Tools
