Did Meta actually auto-enroll brands into AI tools without consent?

Yes — and this is well-documented. Meta's Advantage+ creative enhancements are opt-out by default, not opt-in. REI's deformed bicycle ad became the most visible example of what happens when AI creative tools run without explicit brand approval. The settings are buried inside campaign configurations, and Meta has been expanding what's enabled by default over time. If you haven't done a settings audit recently, there's a good chance features you didn't consciously choose are currently running on your account. I'd check Advantage+ creative enhancement settings in every active campaign today.

How do I check if Meta's AI creative tools are running on my account?

Go into Meta Ads Manager and open any active Advantage+ Shopping or standard campaign. Under the ad set or ad level, look for 'Advantage+ creative' or 'Enhance your creative' settings. You'll see toggles for things like image brightness, background generation, text variations, and — most critically — image generation. Turn off anything that involves structural changes to your creative. Keep the lightweight stuff if you want, but image generation and text overlay should be manually reviewed before you let them run. Do this for every campaign, not just one. Meta doesn't apply changes globally.

Is this only a Meta problem, or are Google and Amazon doing the same thing?

All three platforms are moving in this direction. Google's Performance Max creates asset combinations from your uploaded elements, and 'automatically created assets' can pull copy from your website without you approving each variant. Amazon's AI-generated product imagery has been expanding quietly, and their DSP is using purchase signal data in increasingly opaque ways. The REI story got attention because the output was visually obvious — a broken bicycle. But messaging drift from AI-generated text variants on Google is just as real and far harder to catch. The defence strategy is the same across all platforms: audit defaults, lock your source creative, and measure outcomes outside the platform dashboard.

Should DTC brands stop using Meta's AI tools altogether?

No — that's overcorrecting. Some Advantage+ features genuinely improve performance. Advantage+ Shopping Campaigns often outperform manual campaigns on pure ROAS for catalogue-heavy brands. The issue isn't AI tools; it's passive AI adoption — letting defaults run without understanding what they're doing to your creative and your brand. Use AI features deliberately. Opt in to the ones you've tested and trust. Opt out of the ones that touch visual creative or copy without your review. The brands getting this wrong aren't using AI — they're not watching what the AI is doing on their behalf.

How important is owned channel investment as a hedge against platform AI risk?

It's the most important hedge available to DTC brands right now. Email and SMS are the only performance channels where your creative runs exactly as you built it, with no platform AI remixing it before it reaches your customer. Brands that have built strong email programs — with solid flows, healthy list hygiene, and regular campaigns — are genuinely less exposed to platform volatility. If Meta changes its defaults overnight or a new AI feature breaks your visual identity, your email revenue keeps running. That's not a nice-to-have; it's a structural business advantage. Email should be driving 25-40% of revenue for most DTC brands. If yours is below that, the gap is worth closing.

What's the first thing a DTC brand should do after reading this?

Run a 20-minute Meta settings audit today. Open every active campaign in Ads Manager and check Advantage+ creative enhancement settings. Write down what's enabled and what you're turning off. Then do the same for Google — check your automatically created assets settings in every Performance Max campaign. That's the immediate action. The longer-term work is building a brand control system: documented guidelines, a monthly platform audit workflow, and a measurement setup that doesn't rely solely on in-platform data. If you're not sure where to start, this is exactly the kind of audit we do during paid media onboarding at GOSH Digital.

Meta's AI Auto-Enrollment Is a Brand Control Crisis

REI ran a deformed bicycle ad on Meta. They didn't approve it. They didn't even know it was live.

That's not a hypothetical horror story — it's what Cannes Lions week revealed about where platform AI is heading, and it should make every DTC brand founder deeply uncomfortable.

What Actually Happened With REI (And Why It's Bigger Than One Bad Ad)

Meta's AI creative tools auto-enrolled REI into Advantage+ creative enhancements — without explicit opt-in. The result was a warped, AI-generated bicycle image that ran as a paid ad under REI's brand name. When the story surfaced at Cannes, Meta's response was essentially: this is the direction we're going.

That's the part most trade coverage glossed over.

This wasn't a bug. It was a feature behaving exactly as designed — Meta's AI generating and serving creative variations autonomously, with brands discovering it after the fact. The "opt-out if you don't like it" framing puts the compliance burden entirely on you, while Meta defaults to maximum AI control.

I've been watching this pattern develop across the 40+ brands we manage at GOSH Digital. The shift is real, and it's accelerating.

Platform AI Isn't Your Partner — It's Your Landlord

Here's the uncomfortable truth most paid media content won't say directly: Meta, Google, and Amazon are not building AI tools to help your brand. They're building AI tools to help themselves extract more ad revenue from their inventory.

Your brand guidelines, your creative direction, your carefully built visual identity — none of that is a priority in an automated auction that optimises for clicks.

We've seen this with Google Performance Max, where asset groups get remixed in ways that would make your creative director flinch. We've seen it in Meta Ads campaigns where Advantage+ Shopping quietly de-prioritises your best-performing manual placements in favour of its own signal-heavy black box.

The REI situation is just the first time it produced something visually grotesque enough to make the news.

A deformed bicycle is easy to point at. A subtly off-brand headline that erodes your positioning over six months is much harder to detect — and far more damaging.

The Three Layers of Brand Control You're Probably Not Protecting

When we audit new clients, brand control failures almost always fall into three buckets. Most founders are only watching the first one.

1. Visual Creative Integrity

This is what the REI story is about. AI image generation and creative enhancement tools are now embedded at the platform level — not just in the tools you choose to use. Meta's Advantage+ enhancements, Google's automatically created assets, Amazon's AI-generated product imagery: all of these can run without you signing off on each execution.

The fix isn't complicated, but it requires discipline. You need to explicitly audit your Advantage+ settings inside Meta Ads Manager right now. Go to your campaign settings and check whether "Enhance your creative" is enabled. For most DTC brands, I'd recommend turning off image generation enhancements and keeping only the lightweight stuff like brightness/contrast adjustments — not structural changes to your creative.

2. Messaging and Positioning Drift

Platform AI doesn't understand your brand story. It understands engagement signals.

If a headline that slightly misrepresents your product gets a higher CTR, the algorithm will serve it more. It doesn't know that the headline is misleading — it just knows it converts at the top of the funnel. The downstream effect is returns, support tickets, and LTV destruction that never shows up in your ROAS dashboard.

This is why ad creative measurement has to go beyond platform-reported metrics. You need to track post-click behaviour, return rates by creative variant, and customer satisfaction scores tied to acquisition cohorts. The platform will never surface this data for you voluntarily.

3. Data Sovereignty

The Cannes week coverage also flagged Amazon's expanding commerce data footprint — Amazon DSP is now pulling purchase signal data in ways that make the ad targeting increasingly opaque to brands advertising on the platform.

This connects directly to a broader problem: eCommerce data privacy and data strategy are being outpaced by platform capability. Your customers' purchase behaviour is being used to train models that you have no visibility into and no control over.

If you're not building your own customer data strategy with first-party data at the centre, you're building your business on infrastructure you don't own — and that can change its terms at any time.

What Cannes Actually Revealed (That Nobody's Saying Loudly)

Cannes Lions used to be about creativity. This year it was about control — specifically, who has it.

The £1.6bn UK TV merger in the same week's news cycle is part of the same story. Consolidation at the platform and media owner level always flows in one direction: fewer choices for brands, more power for the platforms. The new shared language for programmatic auctions that also surfaced that week? That's about standardising a system where your creative gets scored, ranked, and potentially modified before it reaches a human eye.

I'm not arguing you should pull spend from Meta. That would be bad advice for most DTC brands at growth stage. I'm arguing that you need to treat platform AI like a contractor who doesn't share your values — useful, but requiring supervision.

The brands that will win this decade are the ones that use platform AI for scale and efficiency while maintaining creative and data control outside the platform walls.

The Practical Defence Playbook for DTC Brands

This is what we actually do for clients at GOSH Digital when we're managing Meta and Google accounts with brand control in mind.

Lock your creative at the source. Upload finished, production-ready assets rather than giving platforms raw elements to recombine. The more complete your creative input, the less room the algorithm has to "enhance" it. Use specific naming conventions that signal to your team (and to audits) which assets have been approved for AI remixing and which haven't.

Audit Advantage+ settings monthly. Meta updates what's opted-in by default more frequently than most brands check. Build a monthly settings audit into your paid media workflow. It takes 20 minutes and has saved our clients from exactly the kind of situation REI found themselves in.

Run a parallel owned channel. Your Klaviyo flows and email channel are the one place where your creative is never remixed by an algorithm without your knowledge. Email revenue as a percentage of total revenue should be climbing for DTC brands right now — partly because it's the last channel you fully control. If you're not investing here, you're more exposed to platform risk than you realise.

Build creative that's AI-resistant by design. This sounds counter-intuitive, but brand-specific creative — real photography of real people using your actual product, with distinctive visual language — is harder for AI to remix convincingly than generic lifestyle imagery. The more generic your creative library, the more vulnerable you are to platform AI that can substitute similar-looking assets.

Know your numbers independent of platform attribution. If your only view of performance is inside Meta Ads Manager or Google Ads, you're looking at data that's been curated to show the platform in a good light. eCommerce analytics that sit outside the ad platforms — MTA models, post-purchase surveys, incrementality testing — give you the signal you need to push back when platform AI makes bad calls.

The Brands That Are Getting This Right

The DTC brands I respect most right now are the ones treating their brand identity like infrastructure — not decoration.

They've got documented brand guidelines that their agency, their internal team, and their platform reps all have access to. They've got a clear list of what can and can't be modified by platform AI tools. And they've got a review cadence that catches drift before it becomes a deformed bicycle.

This isn't complicated. It's just disciplined. And it's the kind of discipline that your marketing stack needs to encode systematically, not leave to individual account managers remembering to check a setting.

The AI in eCommerce marketing conversation has been too optimistic for too long. Yes, AI tools can write product descriptions faster, personalise email content, and optimise bid strategies. We use them constantly.

But there's a version of AI adoption that's passive — where you let platforms make decisions by default and only notice when something goes visibly wrong. That's the version REI got caught by. It's also the version that slowly erodes what makes your brand worth advertising in the first place.

One More Thing the REI Story Gets Right

The reaction to the deformed bicycle ad shouldn't just be "Meta needs better guardrails." The more important lesson is: REI should have had their own guardrails in place before Meta needed to build any.

Platform trust is not a brand strategy.

If you're spending serious money on paid media — whether that's $5k/month or $500k/month — the sophistication of your brand control systems needs to match the sophistication of your spend. The platforms are moving fast. They're not waiting for your approval. That's not a complaint; it's just the reality you're operating in.

Build your brand like you own it. Because on your owned channels, your product pages, your email flows, your packaging — you do. On Meta's platform, you're a tenant. Act accordingly.

If you want to audit your current Meta and Google setup for AI auto-enrollment risks, we do this as part of our paid media onboarding at GOSH Digital. It takes a week, and what we find usually surprises founders who've been running accounts for years.