Brainlabs Managing Director Liz DeAngelis warned on 9 November 2026 that Meta’s Muse Image AI-generated creative could improve campaign results but leave brands unsure which elements drove success. The agency says the lack of clear insight threatens the ability to apply learnings across other channels, making controlled testing essential.
What happened
- Brainlabs published a column titled “If Meta Keeps Optimizing Your Creative, What Are You Actually Learning?” on 9 November 2026.
- The piece focuses on Meta’s Muse Image model, which launched on 7 July 2026 and was promised to be available in Advantage+ creative within weeks.
- According to Brainlabs, 125 days have passed since launch and the integration is still pending.
- DeAngelis argues that letting Meta generate and continuously tweak ad assets can improve performance metrics while hiding which image, message, format or CTA caused the lift.
- She notes that traditional creative testing isolates one variable at a time and carries the insight to future briefs, a loop that AI-driven continuous variation can break.
- Muse Image is priced at $0.01 per image and is listed for ad variant generation, catalog imagery and per-user personalization on Meta’s developer page.
Why it matters
- Brands rely on creative learnings from Facebook and Instagram to inform campaigns on TikTok, YouTube and other platforms.
- If those learnings stay inside Meta’s walled garden, marketers lose the ability to transfer insights to external creative teams.
- Continuous AI-driven changes reduce the chance to isolate the “thing that worked” and turn performance gains into a black box.
- Without detailed reporting on each variation, advertisers may attribute success to Meta rather than to specific creative elements they can replicate elsewhere.
- The risk extends to budget allocation, audience targeting and long-term brand strategy, all of which benefit from transparent testing.
What this means for marketers
- Test AI-generated creative on a small, defined slice of traffic rather than the full campaign.
- Keep a control group that uses manually produced assets to compare performance and isolate variables.
- Capture detailed logs of each AI variation, including image version, copy changes and placement, to feed data back into your own analytics stack.
- Consider using autonomous SEO and GEO services to surface learnings outside Meta’s platform, ensuring insights are searchable and cited in AI answers.
- Maintain relationships with external AI and creative partners to retain control over generation and reporting, as Brainlabs recommends.
- By keeping generation arm’s length from media buying, brands can better understand what resonates and apply those lessons across TikTok, YouTube and other channels.
"Meta is bringing in a landscaper to their walled garden" Liz DeAngelis, Managing Director, Programmatic & Media
The bottom line
Meta’s Muse Image promises faster, cheaper creative production, but Brainlabs warns it can also obscure the very insights marketers need to grow. Controlled testing, clear reporting and keeping AI generation outside the media platform help preserve learnings and protect brand strategy. Brands that act now can enjoy the efficiency of AI while still knowing what truly drives performance.
Sources
- 1.ppc.land