Fashion just split in half over AI, and there's no middle ground.

September 2025. New York Fashion Week hosted two events the same week. One celebrated AI's creative potential. The other documented systematic job elimination. Same city. Completely different movies.

On one side: legendary designers building digital empires. On the other: workers losing livelihoods without consent or compensation. The gap between them is widening every quarter.

The uncomfortable truth? Both sides are right. And the brands choosing sides right now are either building defensible advantages or expensive lawsuits.

New York just passed the first law. Other fashion capitals will follow.

Inside this issue:

✅ Cover Story: Fashion's AI Civil War — Why the industry can't agree on its future
✅ Swipe File: How one TikTok generated $1.25M in media value
✅ Spotlight: ChatGPT 5.1 — Brand voice at scale
✅ Fun Fact: The designer who started at 41

🕒 Read time: 5 minutes

Fashion's AI Civil War: Models Lose Jobs While Legends Build Digital Empires

The industry is splitting into two camps with zero middle ground

September 2025. New York Fashion Week. Two things happened at once.

Brands showcased AI collections and digital model clones. Down the street, Cornell researchers presented evidence that AI is systematically eliminating modelling careers.

Same week. Same city. Different movies.

The divide

In April 2025, 80-year-old fashion icon Norma Kamali announced her proprietary AI trained on 57 years of design archives. She completed MIT's AI course in 2023. Her pitch: "AI could be my Karl Lagerfeld."

In June 2025, New York's Fashion Workers Act took effect,the first law protecting models from unauthorized AI replicas.

Same month: H&M announced plans to create 30 digital model twins. Models would own rights to their avatars and get paid per use.

The tension is operational, not theoretical.

The numbers

AI fashion market: $1.26B (2024) → $1.77B (2025). That's 40.4% growth in one year.

McKinsey's projection: AI will add $150-275 billion to fashion profits in 3-5 years.

50% of fashion executives say product discovery is their top AI use case for 2025.

What's actually happening

The believer camp: Paris-based Heuritech analyzes social media and runways for Prada, Skims, and New Balance. They predicted dotted prints, flat sandals, and yellow for 2025 months before mainstream adoption.

Kamali's AI reimagines her signature pieces in infinite variations. Even the glitches inspire her: "Some of the best editorial fashion is absurd."

The worker reality: Cornell's Worker Institute documented brands hiring contractors to use AI to turn one photo shoot into dozens of images with different outfits. One payment. No additional compensation.

Models can't unionize they're independent contractors. Agencies sign contracts for them.

What the law says

The Fashion Workers Act requires written consent before creating digital replicas. Modeling agencies must register with New York State. Contract transparency is mandatory.

It's the first labor protection law specifically addressing AI displacement in creative work.

Other fashion capitals will copy this template.

Why this matters

Fashion is building 10-100x cost advantages with AI. But unlike Hollywood actors who negotiated AI protections through unions, fashion workers have zero collective bargaining power.

H&M is testing consent-based digital twins with revenue sharing. That's the compliant path.

Most brands are not doing that.

The brands moving without consent protocols are building legal and reputational liability.

What to do

Students: Learn Midjourney, Runway, and generative fashion tools. But also study the Fashion Workers Act. Brands need people who can build AI workflows without creating lawsuits. That's your edge.

Professionals: Pitch AI pilots with consent protocols built in. Frame it as "AI amplification with worker protection." The people building ethical AI workflows now will run departments in 18 months.

Brands: Audit your AI usage against the Fashion Workers Act ,even if you're not in New York. Do you have explicit consent for digital replicas? Are models compensated for AI variations? If not, you're building exposure. California and EU fashion capitals will legislate next.

The playbook

Map legal exposure first. If you modify model images or generate digital replicas, audit consent documentation.

Differentiate by risk. Trend forecasting and product photography = low risk. Digital human generation = maximum risk. Start with high-value, low-risk.

Build consent before scaling. H&M's 30 digital twins have clear ownership rights for models. Operationally harder. Legally defensible.

Watch the timeline. New York legislated first. Other markets follow.

The bottom line

AI trend forecasting works. Heuritech's predictions for 2025 were accurate.

AI job displacement is real. Brands are multiplying assets without additional model payment.

Legislative response is happening. The Fashion Workers Act is operational.

The question isn't whether fashion adopts AI. It's whether fashion builds systems that create lawsuits and backlash—or frameworks that scale with consent built in.

The gap between these approaches is widening every quarter.

Swipe File (Real tactics used by real brands)

🍞 Marc Jacobs' "Make It From Scratch" Play

What they did: Marc Jacobs partnered with TikTok creator Nara Smith (8.7M followers) to promote their tote bag in July 2024. Smith is famous for making everything "from scratch"—including bubblegum and Coca-Cola. So Marc Jacobs leaned in. The video shows Smith mixing flour, eggs, water, and red food dye to "bake" a Marc Jacobs tote bag from scratch. She kneads the dough, rolls it out, puts it in the oven, and pulls out a full-sized red tote bag.

Why it works:

• Format respect: Marc Jacobs adapted their message to fit Smith's signature style. The campaign feels native to her content, not like an ad.

• Measurable explosion: 6.1 million views and 1.2 million likes—Marc Jacobs' highest-performing video at the time. Generated $966,000 in media value on TikTok and $285,000 on Instagram in 48 hours. Total: $1.25M.

• Self-aware absurdity: Viewers know you can't bake a leather bag. That's the point. Comments: "This actually made me want to buy the bag."

Lesson for founders & brand builders: Stop forcing influencers into your brand guidelines. Find creators whose natural format aligns with your message, then get out of their way. Marc Jacobs gave Smith a tote bag and trusted her process. Result: $1.25M in media value for a single TikTok. When format is native and creator is authentic, audiences buy.

The Runway Reel

View more on Instagram

Spotlight

This week’s pick:
This week's pick: ChatGPT 5.1 , Brand Voice Personalization at Scale

Yesterday, OpenAI launched ChatGPT 5.1 with eight personality presets and granular tone controls—the first time an AI model lets brands dial in voice consistency across thousands of interactions.

Why it matters:

The update introduces two models: GPT-5.1 Instant (warmer, conversational) and GPT-5.1 Thinking (deep reasoning). But the game-changer is personality customization. Brands can now set ChatGPT to speak as "Professional," "Candid," "Quirky," "Friendly," or five other preset voices—and fine-tune warmth, conciseness, and even emoji usage.

According to OpenAI, GPT-5.1 is "not only smart, but also enjoyable to talk to," with testers describing responses as surprisingly "playful." The model is also significantly better at following specific instructions—critical for maintaining brand guidelines at scale.

What this means for brands:

• Voice consistency across touchpoints: Customer service, social responses, email campaigns, product descriptions—all can maintain the same brand personality without manual oversight.

• Creative execution at speed: Fashion brands can generate on-brand captions, product copy, and campaign concepts in seconds, with tone locked to brand identity. No more generic AI voice.

• Personalization without fragmentation: Different departments can use the same AI but with different personality settings. Marketing gets "Quirky," customer service gets "Professional," creative gets "Candid."

The takeaway:

Generic AI voice is dead. ChatGPT 5.1 makes brand personality programmable. Fashion brands that nail their voice settings now will produce 10x more content without losing brand identity. The ones still manually writing every caption are falling behind.

A Final Note

Fun Fact

Giorgio Armani founded his fashion empire at 41 after abandoning a career in medicine. He studied at the University of Milan's Medicine Department before leaving to serve in the Italian Army. After military service, he worked as a window dresser at La Rinascente department store, then spent years designing for Nino Cerruti. In 1975, he and partner Sergio Galeotti founded Giorgio Armani S.p.A. with seed money from selling Armani's old VW Beetle. By the 1980s, his "power suits" became iconic through American Gigolo and Miami Vice.

Let me know what you thought of this edition.Until next time,