If you’re reading this, you already know the game has changed.
We’re still in the Wild West of AI deployment — and this issue is a field guide to what happens when nobody’s holding the reins.
Every story below traces back to the same root cause: AI systems left to run without checking the work: a swapped-out ad (creative inaccuracy), a coupon code gone rogue (financial loss), a chatbot with bad legal advice (legal exposure), a reputation that won't die (reputation damage), and AI-sounding copy nobody caught (authenticity erosion).
I learned a lesson early on in business: “People do what you inspect, not what you expect.” This is a truth for most human workers, and it’s proving even more true of AI bots and apps. Here are somethings that went wrong, who it happened to, and exactly what to check before it happens to you.
Welcome to LRND™ Intelligence. Let’s get into it.
P.S. Scroll down for a Partner Offer from a Predictive AI Intelligence Firm to the first 5 respondents. 👇
True Classic’s marketing chief discovered that Meta’s ad platform had autonomously replaced the brand’s top-performing ad — a casual shot of a man in a matching fleece set — with an AI-generated version featuring an unapproved, unrelated character. It wasn’t a one-off glitch: Meta’s Advantage+ platform has been swapping advertisers’ top-performing ads for AI-generated alternatives without prior approval as a matter of course.
What it means for you: Your “top-performing ad” isn’t necessarily the ad running right now. Check your Advantage+ account settings this week — the auto-creative-testing toggle is often on by default, and most marketers don’t discover it until their winning creative has already been replaced.
One company’s support chatbot applied the same promotional code repeatedly to a single order, generating negative prices where the business was effectively paying customers — resulting in over $150,000 in losses across 2,400 fraudulent orders before anyone caught it.
What it means for you: This isn’t a brand-safety review problem, it’s a finance problem. Reputational monitoring won’t catch a bot quietly bleeding margin one transaction at a time — someone on your team needs real-time anomaly alerts on anything your AI can apply to price.
Coverage of a past Southwest crisis continues to surface in the airline’s branded AI search results in 2026, showing that generative engines keep resurfacing outdated negative narratives long after the news cycle has moved on.
What it means for you: AI answer engines don’t “move on” the way search rankings or the news cycle do — a bad week from years ago can sit permanently in the first thing a customer sees when they ask an AI about you. Reputation management now needs an active, ongoing content strategy aimed specifically at what LLMs cite, not just what ranks on page one of Google.
PARTNER OFFER
Our friends at Sooth® THE Predictive AI Intelligence Firm are offering a complimentary ’Snapshot’ to the first five respondents with $10M+ revenue doing business in the US.
The Sooth engine, ELI (Emotional Logic Interface) reads the real emotions behind people’s choices and scores what drives action - trained on 100M+ verified signals, 91% accurate. The result: a Sooth Snapshot, a one-page read on any audience showing where your message lands, where it misses, and the one change that moves the needle.
New York City’s “MyCity” business chatbot was found telling employers they could legally take a cut of workers’ tips and telling users that landlords didn’t have to accept housing vouchers — both incorrect, the second a discrimination violation — and the city pulled it offline, with officials calling it “functionally unusable.”
What it means for you: A chatbot doesn’t need a lawsuit to become a liability — one bad screenshot of confidently wrong advice can end the product. If your bot answers any question touching legal, financial, or policy territory, that content needs the same review rigor as anything a human rep is trained and scripted to say.
Someone spotted what looked like AI-written copy on a Nike product page and posted it publicly, asking why a marketing team would let an obvious “GPT AI-ism” through on the brand’s main page. Fast Company covered the moment, noting brands are now “particularly vulnerable to charges of inauthenticity,” with AI acting as an “inauthenticity force multiplier” — and Nike never ran a deliberate AI campaign at all. This wasn’t a splashy ad campaign gone wrong. It was routine product copy that slipped past review.
What it means for you: Audiences aren’t just reacting to AI mistakes anymore — they’re pattern-matching for AI tone, even in the most boring corners of your site. A product description doesn’t get the same scrutiny as a hero ad, but it’s just as exposed. If your copy pipeline touches AI anywhere, your review process needs to cover the unglamorous, everyday pages too — not just the campaigns everyone’s already watching closely.
Thanks for reading Issue No. 09 of LRND™ Intelligence. We’ll be back with more of what matters — no filler, no fluff.
→ Want to talk about your brand’s strategy for 2026? Let’s connect.
LRND™ is an AI-powered, independent creative, development, and media agency. We employ equal parts man and machine to deliver work that works.
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.