Julie Bornstein's Daydream has the Forbes profile. Our full piece is out today — read that for the real argument. The short version: Bornstein bet on curatorial judgment rather than checkout conversion, which is either the right instinct or a very expensive one, depending on what Sephora does in the next six months. See below.

Sephora is plugging into Google's AI shopping ecosystem — beauty advisors, product discovery, loyalty personalization. This is the third major retailer this year to make this exact move. John Lewis went with ChatGPT and Google in May. OTB put five luxury houses on a single Google Cloud AI layer. The pattern is structural: mid-to-large retailers are not building their own AI stacks. They are renting Google's, and calling it a strategy.

What Sephora gives up in this deal is the question nobody's asking. Beauty Insider has 34 million members. That loyalty data, fed into Google's shopping layer, helps Google more than it helps Sephora. The differentiation Sephora is known for — product discovery, advisor expertise — is exactly what Google is trying to replicate for every retailer simultaneously. You can read this as Sephora getting smarter, or as Sephora paying Google to make itself a commodity. Both readings are available from the same press release.

亿邦动力网 reports Pinduoduo is tightening AI content governance — stricter thresholds on AI-generated product copy, images, promotional text. This is not a values statement. At Pinduoduo's scale, AI-generated slop is a search quality problem, which is a GMV problem. They're using AI to police AI content because the alternative is watching the catalog degrade faster than any human moderation team can touch.

Cross this with the German fraud-prevention piece below and the shape becomes clear: AI is simultaneously the content creator, the content regulator, the shopper, and the fraud vector. The platform managing all four roles at once is running the most complicated conflict-of-interest structure in commerce history, and the conflict is only getting worse.

Futu — Hong Kong's brokerage platform, not a trade publication — asks whether Pinduoduo's model is vulnerable to AI disruption. The bear case: AI competitors can replicate Pinduoduo's price aggregation and targeting without the supply chain years. The bull case: Pinduoduo's real moat is logistics density and merchant lock-in, neither of which AI replaces overnight. The honest answer is that the squeeze is slower than the headline implies. What matters is that Futu is asking the question seriously at all. Six months ago, nobody was.

证券时报 has the forensics: JD is poaching Tmall merchants, Pinduoduo is winning on price, Douyin is eating the discovery layer. Taobao's two engines — Tmall premium and Taobao mass market — are stalling simultaneously while Alibaba commits ¥380 billion to AI infrastructure. We flagged this defensive pattern when the pledge was announced. What the Securities Times adds is pace: the capex story and the GMV story are moving in opposite directions at the same time. That is not the position Alibaba's board designed for.

Pascal Bohlmann in etailment asks the question every fraud team is quietly panicking about: at checkout, is the entity completing the purchase a human, an authorized AI agent, or an attacker mimicking one? From the outside, all three are operationally identical. Current fraud models run on human behavioral signals — mouse dynamics, session timing, keystroke patterns. An AI agent produces none of these, which means it simultaneously triggers no flags and leaves no fingerprint.

Neritus Vale's piece today — The Shopping Agent Has to See Your Screen. That's the Problem. — covers the permission side of agentic commerce. The fraud problem is the other face of the same coin: even if you solve for what the agent is allowed to do, you haven't solved for who sent it. These are different problems and right now, no one has a working answer to either.

FashionUnited has a competent overview of recent UK copycat case law — the UGG slipper ruling, the general pattern of inconsistent protection. Sir John Crabstone has the full argument out today: AI Clones a Design Overnight. UK Copycat Law Still Argues Over Handbags. Go there for the actual stakes. The short version: UK design protection moves in years, AI clone cycles now move in hours, and the math does not work in the claimant's favor at any point in that timeline.

FashionUnited has brands talking on record about DPP implementation under the Ecodesign Regulation. The ones willing to speak publicly are the ones that started two years ago; everyone else is still arguing internally over who owns the data infrastructure. The compliance framing misses the actual asset being built: a DPP is a product-level API. When every garment carries a machine-readable chain of custody, that data becomes usable for resale authentication, AI-powered sourcing decisions, and item-level traceability that makes counterfeiting structurally harder over time.

Worth pairing with our piece on how the counterfeit supply chain had already built its own compliance records — clean on paper while the cargo moved elsewhere. The DPP assumes good-faith actors hold the clean records. That assumption has a documented failure mode.

Hillary Super took the Victoria's Secret CEO role in March 2024, deflected an activist board challenge with nine days to spare before the shareholder vote, and is now posting what FashionUnited is calling a record turnaround. CEO profiles carry their own editorial bias, but the share price confirms the narrative, which makes it harder to dismiss. Super has been specific that AI-assisted sizing and fit personalization is central to the recovery thesis — which is the correct diagnosis. Victoria's Secret's core product problem was always the gap between catalog fantasy and physical fit reality. Virtual try-on never closed that gap. AI-assisted sizing at scale can, and this is apparently where the technology investment actually went.

Topshop paired an immersive catwalk show with a TikTok live beauty shopping experience. The architecture is the point: the catwalk provides cultural legitimacy, the TikTok live provides the conversion mechanism, and neither works alone. This is essentially the TikTok Shop format running on a British heritage brand. Relevant that TikTok Shop's 84% beauty surge last year ran through comment-thread conversion rather than creator video views — Topshop is plugging into that same mechanic. Whether a brand with this much archive nostalgia can sustain the content cadence live commerce demands is the operational question the press release does not answer.

BoF's opinion piece frames AI shopping as a coming threat to the "existing online shopping experience," with retailers at risk of being left behind if they don't move fast enough. The framing is backwards. AI isn't killing online shopping — it's moving the interface one layer up, from the retailer's own site into the agent's conversation layer. The question that actually matters is not whether shopping survives but who ends up inside the agent's recommendation index and who gets cut out, the way comparison-shopping engines cut out direct booking a decade ago.

Right now the answer to "who controls the recommendations" is whoever signed a deal with Google last week. See Sephora, above. Framing this as a technology adoption problem — "will retailers get left behind?" — misses where the power is going. It's a distribution concentration problem, and it is concentrating exactly as you would expect.

Laura du Rusquec — most recently at Ganni — takes the CEO chair at Kylie Jenner's Khy. The Ganni credential carries weight: that brand ran a genuine DTC-to-wholesale pivot, built supply chain infrastructure for sustainability commitments it actually filed paperwork on, and navigated fast-fashion adjacency without collapsing its brand equity. The question Khy has never answered — fashion brand or well-photographed influencer merch — is now being answered by the hire. Whether the operational infrastructure follows is the test.

A foraged-fibre collection at New Designers 2026 — garments woven to respond to touch, light, and movement. Graduate work, not commercial release, and the caveats are the usual ones. But the materials direction is worth tracking regardless: the push toward foraged and living inputs is happening at every tier simultaneously, from arthouse graduate shows to Tier 1 sourcing briefs. The commercial distance is long. The vocabulary is forming now, and the brands that ignore graduate materials work are the ones surprised when it turns up in a competitor's sourcing deck three years later.