Issue
No. 03.

This Week in AI + Advertising

1. 1

   You Can Now Buy From D2C Brands Without Ever Leaving ChatGPT.

   GoKwik and PayU just launched India’s first multi-brand shopping trip that never leaves a chat window - discover, compare, and check out via UPI or card, no tabs required. Hyphen, Beardo, and Kilrr Spices are already live, with hundreds more from GoKwik’s network lining up behind them. Forget “add to cart.” This is “add to conversation.” If we’ve got a D2C client, their next distribution channel might genuinely be a chatbot.

2. 2

   WPP Just Handed an AI Agent the Keys to Its Ad Budget.

   WPP Media partnered with Kily - an agentic AI startup built by ex-Flipkart and Affle folks - to let AI agents autonomously handle listing, pricing, and ad decisions across India’s quick-commerce chaos: Blinkit, Zepto, the works. No human in the loop for the boring stuff. Same story as #1: the AI isn’t just recommending anymore, it’s doing. And it’s happening in our backyard first.

3. 3

   70% of Teams “Use AI.” 88% Still Need a Human to Clean Up After It.

   Knak asked 333 enterprise marketers the real question - not “do you use AI” but “does its work actually ship as-is.” Turns out: nope. One AI-assisted email still needs 3-5 tools, 2-3 rounds of human editing, and roughly $300 in labor before it’s fit to send. AI got everyone to a first draft. Nobody told it that’s not the same as being done. Keep this one in your pocket for the next “why can’t AI just do it” conversation.

4. 4

   A Brand Used AI to Pose a Model the Way She Never Actually Stood. She’s Suing.

   Creator Francheska Pujols is suing clothing brand Rainbow Shops, alleging it used AI to generate images of her in poses she never struck during the shoot - including one straddling a barstool - then ran them as if she had. “I felt disrespected and violated,” she told Bloomberg Law. She’s not alone: brands are quietly using AI to “fix” or extend creator content after the shoot wraps, and creators are finding out when the ad goes live. If a client ever wants to “enhance” someone’s content with AI, get it in writing first. This is what happens when nobody does.

5. 5

   Europe Just Told Every Chatbot to Introduce Itself.

   As of yesterday, EU law requires AI systems to say “hi, I’m a robot” before things go too far - chatbots must disclose, deepfakes need labels, AI content has to be machine-detectable. Fines top out at €15 million or 3% of global revenue, whichever hurts more, and it applies to anyone whose content reaches EU eyeballs, not just EU companies. The single biggest compliance shift of the week, even without a euro of product on our books.