L'Oréal CMO Asmita Dubey has extended the company's OpenAI relationship to accelerate the generative AI content engine the brand has been building for two years — the same engine we wrote about as a machine producing content nobody clicks. The fresh deal is either conviction that the production-cost collapse justifies the investment even if the distribution problem remains unsolved, or it's optics: Cannes Lions this month made AI the industry's credibility flex, and L'Oréal needs to stay ahead of a pack that has now caught up.

The deeper problem is structural. L'Oréal is producing more content for consumers who are reading less content. Every asset from a generative pipeline competes with every other brand running the same pipeline on the same platforms. When everyone accelerates, the advantage goes to whoever has better targeting — and targeting is a data game, not a creative one. Dubey is smart enough to know this. Which means either the OpenAI deal is about something we're not being told, or L'Oréal is buying time while hoping AI makes content cheap enough to survive the zero-click transition regardless.

Julie Bornstein's Daydream — an AI personal stylist for fashion discovery — is getting the full founder-visionary treatment in Forbes. The thesis is real: better discovery should reduce the mismatch between what people order and what they keep. But shopping agents score 76% on a simulated customer, not a real one, and fashion returns have always been more about fit, material feel, and purchase anxiety than bad search. Daydream solves the intent layer. The body, the fabric, the light in your bedroom — still not in the model.

Chinese retail trade site eBrun reports Pinduoduo is tightening compliance thresholds for AI-generated listings and creator content — raising the governance bar precisely as Western brands race to automate their creative pipelines. The timing is not coincidental. Alibaba is spending ¥380 billion on AI infrastructure to catch its rivals; Pinduoduo is simultaneously making the content environment on its own platform harder for cheap AI-generated noise to pollute. Every platform moving product at scale is now running both sides of this equation: accelerate AI adoption, then regulate the waste it generates. The West will arrive at this same moment, just later.

Securities Times details how Taobao-Tmall's growth is decelerating as JD poaches merchants on reliability, Pinduoduo owns the value floor, and Douyin absorbs discovery. The ¥380 billion AI pledge reads, correctly, as defensive capex. Read this alongside our earlier piece: Alibaba's bet is on the story that still grows, not the commerce that already isn't. The question remains whether infrastructure spend translates to recovered commerce share, and the answer keeps being: not automatically, and not quickly.

German trade outlet etailment.de asks the question keeping every risk team up at night: when an AI agent executes a purchase on behalf of a human, how do you distinguish an authorized agent from a bot attack? Current fraud infrastructure — behavioral biometrics, device fingerprinting, CAPTCHA — was built to catch humans pretending to be bots. An authorized AI agent looks like exactly the wrong thing from every detection angle: inhuman speed, unfamiliar IP, scripted checkout flow.

Nobody has written the liability contract for agentic fraud yet. When a compromised AI agent buys €3,000 of goods using stolen credentials, who absorbs the chargeback — the orchestration layer, the wallet provider, the platform? The platforms moving fastest on agentic commerce are also the most exposed when this stops being theoretical. The Business of Fashion asked this week whether AI will be online shopping's next victim; etailment's framing is sharper — the victim might be your customer's bank account.

Kay Barron, former fashion director at NET-A-PORTER, has launched VVEND — a live commerce platform built explicitly for luxury, not clearance. The bet is that live selling has been marooned in the TikTok Shop format: high volume, price-driven, slightly chaotic. Barron's argument is that the editorial authority that made NET-A-PORTER work translates to live: slower, more considered, curator-fronted. TikTok Shop's beauty numbers ran on comments doing the closing — VVEND wants to reverse the polarity, making the host the event, not the comment thread.

Meanwhile, Dove's World Cup creator fleet shows what happens when creator scale outruns brand safety controls — the live format Barron is building has the same tension at much higher stakes per unit sold. Whether luxury customers want to watch someone sell them a bag in real time is the open question. Barron's credibility is the product, for now.

FashionUnited's profile of Jaume Miquel, CEO of Spanish group Tendam (Springfield, Women'secret, Cortefiel), is the best case study in European mid-market fashion data strategy you'll read this month. Ten years of feedback loops between store performance and assortment planning — not at Inditex's unobservable scale, but at a size where you can actually see the architecture. Miquel's specific articulation of how data changed merchandising decisions is worth more than most vendor-sponsored AI transformation whitepapers. This is the template for how mid-market European fashion earns its next decade. Read it before your competitor does.

StockX launched "Listings" — a peer-to-peer feature that lets sellers price their own secondhand sneakers and apparel without waiting for a verified market transaction. This is StockX acknowledging that its authentication-first, bid/ask structure is too rigid for the seller who just wants to move a pair of shoes on their own schedule. It also puts StockX in direct competition with Depop, Vinted, and the resale layers built into every major fashion platform. The authentication credibility is real and valuable. Whether it survives contact with the casual-seller format — where price, not provenance, drives the transaction — is the test.

Shopify's AI product discovery integration — announced alongside Horizon, its new AI-driven design standard — is the quiet consolidation play that Algolia, Constructor, and the broader search middleware cohort have been watching with justified anxiety. When the platform itself becomes the discovery layer, the argument for a third-party search stack gets harder to make at the SMB and mid-market tier. Retailers shipped apps before anyone built the evaluation frameworks; now Shopify is building the framework into the platform itself. The vendors who survive this are the ones with enterprise-scale differentiation substantial enough that the native Shopify version feels like a toy by comparison. Watch which ones get acquired in the next 18 months. Claire's counter-bet on child influencer stores over app-based discovery reads, against this backdrop, as a deliberate rejection of the entire optimization premise — if the platform will eventually commoditize discovery, the differentiator is the experience you can't run through a model.

Osmo, the AI fragrance startup, is putting patented AI-developed scent molecules up for auction at the World Perfumery Congress. The mechanics are genuinely new: Osmo's model identifies fragrance structures that human noses find appealing but that human chemists haven't synthesized, then patents the resulting molecule. Auctioning those molecules is IP-as-commodity for AI-generated olfactory assets — a sentence from a dystopian novel that is apparently Tuesday in 2026.

The business logic is real: a proprietary molecule that performs better than current equivalents has genuine licensing value to fragrance houses. The question is how you price something whose performance is measured subjectively in a market that historically rewards terroir and heritage over optimization. Swatch is currently seeking $170 million from Samsung for watch face designs that exist only on a screen — the IP frontier on AI-generated sensory assets is nowhere near settled, and the auction at the World Perfumery Congress is going to be one of the more interesting price discovery experiments of the year.

Glossy's newsletter asks whether the bare nail moment is a fashion choice or a financial one. The framing misses the more useful question: the signal isn't the nail, it's the appointment it cancels. Elaborate gel and press-on sets built their own consumer habit around a two-hour experience that's hard to justify when household spending tightens — and elasticity on beauty service time runs higher than elasticity on product. The brands with exposure here aren't polish labels. They're nail studios, beauty service booking platforms, and any brand whose distribution strategy depends on in-salon impulse purchases. When the appointment disappears, so does the discovery moment.

Castore has acquired a majority stake in Grenson, the British heritage bootmaker. Neritus Vale's piece today explains what this acquisition is actually about — short version, Castore is building something that looks less like a sportswear brand and more like a data-first brand portfolio, and Grenson is another data-rich asset in that thesis. The FashionUnited report covers the deal structure. Our piece covers the logic. Read ours.