What Advantage+ Creative actually does to your ads
Advantage+ Creative is Meta's automated creative variation layer, built into Ads Manager at the ad level. When it is active, Meta generates multiple versions of your uploaded creative and serves each version to the viewer Meta's system predicts will respond to it best. Your original creative always runs as one of those versions. The others are AI-modified.
As of July 2026, more than 8 million advertisers are using Meta AI creative tools, according to Meta's July 7 announcement. The platform-level adoption figure matters because it reflects how deeply embedded this toolset is in the default campaign creation flow, not a separate opt-in feature a small group of early adopters has chosen.
Meta performance claim (with caveat)
Meta Marketing Science 2025 reports a 12% higher CTR for campaigns using Advantage+ Creative. These are platform-measured figures on Meta's own infrastructure. Novelty-effect decay and the absence of third-party verification mean the figure should be treated as directional, not deterministic.
The enhancements operate at two levels. Standard enhancements are on by default and adjust existing creative: adding music, changing colour treatment, animating static images, repositioning text. Test enhancements are opt-in and go further: they generate net-new creative content, including AI-synthesised backgrounds and AI-extracted copy overlays, without a preview before the ad runs.
The practical consequence: many advertisers have been running campaigns where AI has been modifying their designed creative without reviewing what those modifications look like. The modifications are not random. They are optimised toward engagement signals. But engagement optimisation and brand guideline compliance are not the same objective.
Standard enhancements versus test enhancements
The distinction between standard and test enhancements is operationally significant. Standard enhancements adjust content you uploaded. Test enhancements generate new content. The opt-in threshold is correspondingly higher for test enhancements: you must accept all standard enhancements before you can opt into the test set, and test impressions are capped at under 5% of total delivery.
| Enhancement | What it does | Primary brand risk |
|---|---|---|
| Music | Adds auto-selected music to Reels and Stories ads | Off-brand or tone-inappropriate track selection |
| Image Filter | Adjusts colour temperature, sharpness, vignette | May alter brand colour fidelity in photography-sensitive creative |
| 3D Animation | Converts static image to motion (zoom, parallax) | May feel wrong for conservative or premium brand categories |
| Image Template | Adds a solid-colour text overlay bar at top of image | Conflicts with designed layouts; widely flagged by practitioners as visually poor |
| Visual Touch-Ups | Adjusts aspect ratio, brightness, contrast | Low risk for most advertisers |
| Text Improvements | Swaps text between headline and primary text positions | Copy remains unchanged; positioning risk only |
| Expand Image | Fills placement space with slightly extended image edges | Low risk for most advertisers |
| Relevant Comments | Surfaces one "most relevant" comment directly on the ad | No preview; negative comments can appear without advertiser control |
| Carousel Enhancements | Profile end card, dynamic description, best card first ordering | Low risk; sequence logic may affect story flow in designed carousels |
| Enhancement | What it does | Brand risk level |
|---|---|---|
| Image Expansion (outpainting) | AI generates new pixels around a square image to fill 9:16 vertical placements | Medium: new content is generated, not just cropped |
| Background Generation | AI generates entirely new backgrounds for product images. No preview before live. | High: brand context is invisible until the ad runs |
| Text Extraction Optimization | Extracts keywords from ad copy and displays as text overlays or footers. No preview. | High: extracted phrases may create misleading or non-compliant impressions |
| Expand Video | AI generates new content above and below landscape video to fill 9:16 | Medium: edge content is AI-generated |
No-preview risk
Background Generation and Text Extraction Optimization have no preview mechanism. Your product image gets a new AI-generated background, and copy phrases are displayed as overlays, without any approval step before the ad runs. For regulated industries (financial services, healthcare, insurance) this is a compliance exposure, not just a brand one.
Brand risk rated by enhancement
Not all enhancements carry the same risk profile. The risk depends on three factors: whether AI is generating net-new content versus adjusting existing content; whether a preview is available before the ad runs; and whether the enhancement operates in a category with regulatory copy requirements.
| Enhancement | Default state | Net-new content? | Preview available? | Risk (regulated sector) |
|---|---|---|---|---|
| Background Generation | Opt-in (test) | Yes | No | High |
| Text Extraction Optimization | Opt-in (test) | Yes (overlays) | No | High |
| Music | On by default | Yes (audio) | Partial | Medium |
| Image Template | On by default | No | Yes | Medium |
| 3D Animation | On by default | No | Yes | Medium (premium brands) |
| Relevant Comments | On by default | No (surfaces existing) | No | Medium |
| Image Expansion | Opt-in (test) | Yes (edge pixels) | Limited | Medium |
| Image Filter | On by default | No | Yes | Low |
| Visual Touch-Ups | On by default | No | Yes | Low |
| Text Improvements | On by default | No (repositions only) | Yes | Low |
One enhancement not in the Advantage+ Creative panel itself warrants attention: Optimize Website Destination, found in campaign settings. This feature allows Meta to redirect traffic to pages on the advertiser's website that Meta predicts will perform better than the set destination, without the advertiser seeing or approving the redirect. For advertisers with specific landing page requirements tied to campaign messaging or compliance, this setting should be reviewed independently of the Advantage+ Creative panel.
Muse Image: what was announced and what got pulled
On July 7, 2026, Meta announced Muse Image, built by Meta Superintelligence Labs. The correct product name is Muse Image (not "Meta Muse," which appears in some third-party coverage and is not Meta's term). The announcement came alongside the Cannes Lions period updates and was published on Meta's newsroom.
Muse Image generates images from text prompts using what Meta describes as agentic visual reasoning: the model plans layout, incorporates real-time web context, and blends multiple visual references. Announced capabilities include photo restoration, style transformation, compositing, text rendering in images, and room redesign with shoppable product placement.
Current availability as of the announcement: Meta AI app, Instagram Stories AI effects (30+ effects), WhatsApp in limited rollout. Facebook and Messenger integration was announced as coming later. Advertiser integration via Advantage+ Creative was announced as "in the coming weeks" from July 7, with a Q3 2026 rollout for larger accounts.
Controversy, July 10
Three days after the July 7 announcement, Meta pulled a feature that allowed users to @-mention public Instagram accounts and generate AI images of those individuals without notification. The feature was disabled on July 10 following user feedback about consent and notification expectations. The pullback is relevant for advertisers using Muse Image for creator-adjacent content: the model's boundaries around depicting real individuals are actively being defined, and the policy can change between announcement and rollout.
For brand advertisers, the Muse Image integration into Advantage+ Creative in Q3 2026 means the test enhancement toolset will have access to a substantially more capable image generation model than the current background generation system. The operational implication is the same: no preview before live, and the gap between what you uploaded and what runs in market will potentially be wider.
What to turn off and what to keep
The decision on which enhancements to disable is not universal. It depends on brand category, creative execution style, and regulatory environment. The guidance below reflects the most common risk scenarios.
For most brand advertisers, disable these standard enhancements:
- Image Template. The text overlay bar conflicts with designed ad creative. There is no case where a designed brand ad benefits from a generic text bar added by the system.
- Music. Unless your brand has established audio identity guidelines and you have tested the auto-selection, the risk of an off-tone or mismatched track outweighs the engagement lift claim for Reels placements.
- Relevant Comments. The surface area for a negative comment appearing on a brand ad is real and there is no preview mechanism. Disable unless you have active comment moderation running and have accepted the exposure.
For regulated industries (financial services, healthcare, insurance, pharmaceuticals), additionally review:
- 3D Animation. Converting a regulatory-compliant static image to motion may introduce visual framing that changes how disclosures or product information reads. Review how your required disclosures appear in animated variants.
- Text Improvements. Text repositioning is low risk for most advertisers, but for campaigns where compliance copy placement is part of the approved creative, verify that repositioning variants maintain required layouts.
For test enhancements, the baseline recommendation is:
- Do not opt into Background Generation or Text Extraction Optimization without a process to monitor what is generated in market. If you do opt in, set a monitoring schedule to review active ads within 48 hours of launch and check enhanced variants against brand guidelines.
- Image Expansion carries lower risk than Background Generation for most advertisers and is appropriate for campaigns where asset fill across placements is the priority.
Practical note on performance isolation
Ads Manager does not break down which specific enhancement is driving performance lift or degradation in a given campaign. You cannot determine from the dashboard that Image Template caused CTR to drop or that Music caused a bounce rate increase. The performance measurement is at the ad level, not the enhancement level. If enhancement-level attribution matters for your creative testing programme, the current toolset does not provide it.
AI disclosure: how this connects to Meta's labeling policy
Meta has operated an AI content labeling policy since February 2024, applying "Made with AI" and "AI info" labels to user-uploaded AI-generated content. The policy expanded in scope as part of the July 7, 2026 Muse Image announcement period.
Industry reporting from the same week indicates Meta is extending its AI label policy to cover modifications made by Advantage+ Creative tools themselves, including Background Generation, Image Generation, and Add Animation, as well as third-party AI-generated content detected via C2PA (Coalition for Content Provenance and Authenticity) metadata standards. This would mean AI-modified ad creative carries a disclosure label the same way user-uploaded AI content does.
This is a meaningful shift in scope. If your ad creative was modified by Advantage+ Creative's Background Generation and an AI label appears on the delivered ad, that label was not in your approved creative. Build monitoring of delivered ad creative into your workflow, particularly for high-visibility or brand-sensitive campaigns, so you can identify when and where AI labels are appearing on your ads.
The direction of travel across platforms is consistent. Google expanded AI ad disclosures to Search, YouTube, and Discover on July 9, 2026 (see our coverage: Google Ads AI transparency labels: what the July 2026 update means). Meta is moving in the same direction. Regulatory frameworks in the EU, US, Singapore, Australia, and Canada are all developing mandatory disclosure requirements that will eventually formalise what platforms are doing voluntarily. Advertisers who build disclosure-aware creative workflows now will have less to retrofit when requirements become mandatory.