We track AI and Meta platform developments every week and apply what's relevant directly to client work. Most weeks have one thing worth acting on. This week had three.

New disclosure requirements for US-market AI ads

The New York AI Synthetic Performer Disclosure Law took effect on 9 June 2026. It requires that any AI-generated or synthetic performer — an AI actor, avatar, UGC spokesperson, or digitally created character — carry a conspicuous visible disclosure in the creative itself across every paid channel: Meta, Google, YouTube, TikTok, CTV, and display.

Civil penalties

$1,000 for a first offense. $5,000 per subsequent violation. These apply to any brand running US-distribution campaigns — the law is not limited to NY-based advertisers.

This stacks directly on top of Meta's existing AI disclosure requirement, which already carries real enforcement weight: undisclosed AI content is now the third-largest ad rejection reason on the platform at roughly 14% of all rejections, and violations accumulate as account strikes, not just rejected ads.

The important distinction: Meta's disclosure checkbox covers Meta's requirement. It does not satisfy the NY law. The NY law requires a visible label in the creative itself — "Digitally created" or equivalent — that a viewer can see without accessing ad metadata. Two separate obligations, two separate checks.

We've updated our campaign launch checklist to make both mandatory steps before any ad with AI-generated performers goes live on any US-distribution campaign. The Meta checkbox is one step. The in-creative label is a separate step. Neither substitutes for the other.

14%
of Meta ad rejections are undisclosed AI content
$5K
civil penalty per repeat violation under NY law
Jun 9
NY law in force — not pending, active now

The Advantage+ trap brands walk into when duplicating ads

Since February 2026, Meta has defaulted all Advantage+ creative enhancements to on for new Sales, Leads, and App Promotion campaigns. That's well-documented. The less obvious issue: those enhancements reset to on every time an ad is duplicated — even if you had disabled them.

Meta confirmed this behaviour persists despite a March 2026 preference-persistence fix. The fix addressed new campaigns; it does not cover duplication resets.

In practice, it means: a brand approves an ad. The campaign runs. You duplicate the ad for a new budget period, a new audience test, or a seasonal push. The duplicated version goes live with Meta's AI modifications applied — different crops, different text overlays, potentially different framing than what the brand signed off on. The original ad stays as approved. The duplicate doesn't.

What we do

We check Advantage+ creative enhancements on every launch and after every duplication as an explicit step. It takes under a minute per ad set. For brand-sensitive clients, this is not optional — it's a live risk on every duplicated campaign.

Better product video from more inputs in a single generation

Seedance 2.0 accepts up to 9 images, 3 video clips (15 seconds total), and 3 audio files as inputs in a single generation — 15 mixed references at once. We've been using it as a single-image-to-video tool. The multi-input capability is a meaningful production upgrade that we're now deploying on client work.

For e-commerce briefs, what this changes: a product launch video can now incorporate multiple product angles, a brand audio cue, and a previous approved spot as a style reference — all in one pass. The model maintains consistency across the full clip: same characters, same scene, same text rendering throughout. More reference inputs means tighter brand consistency and fewer generation iterations to get the output right.

Production impact

Previously, achieving brand consistency across a 15-second product video often meant generating multiple separate clips and cutting them together — with the visual inconsistency that creates. Multi-input generation handles this in a single pass. The result is more coherent and faster to produce.

We're benchmarking multi-input generation against single-input on current client briefs. If the output quality holds — and early results suggest it does — this changes what we can promise on e-commerce production timelines and how many client-approval rounds a product video typically requires.

Why we track this weekly

The AI toolset, Meta's policies, and the compliance landscape all move fast enough that a quarterly review misses things that matter. A disclosure requirement that went live nine days ago is already live enforcement. An ad duplication behaviour that's been running since March is already modifying approved creative on active campaigns.

We run this review every week and apply the relevant updates directly to how we set up and manage campaigns. The point isn't to be first with a newsletter about new AI tools. The point is to make sure client campaigns are compliant, current, and running on the best available production stack — without clients having to track any of this themselves.

That's the done-for-you part.