Meta’s Talent Heist + Google’s AI That Can’t Do Pants 🥷👖

Good morning, fellow brainiacs 🤖

Hope your dopamine levels are as optimized as Claude's new emotional support research 🫂. Today, we’ve got a Meta power play, Google’s AI coming to your pocket, and a virtual try-on app that still thinks pants are optional. Let’s get into it ⬇️

👕 “Fit Check? More Like Fit Fail”

Google’s ‘Doppl’ app lets you try on outfits — kinda

Google's cooking up a new app called Doppl, which uses AI to create video clips of you wearing clothes… without ever stepping into a changing room.

📸 Just upload a full-body pic + clothing screenshot, and Doppl generates an animated lookbook of you strutting in digital drip.

😬 There’s a catch though — Doppl has trouble with pants (yep), sometimes conjures fake feet, and mirror selfies make you mysteriously thinner.

⚙️ Unlike older try-on features that only worked in Search, this app supports outfits from anywhere on the web and builds a full-on animation.

Why it matters:
Virtual try-ons are finally getting cinematic… but clearly need a few stitches. Still, Google’s trying to own AI in fashion before TikTok filters beat them to it.

🥊 “Meta Raids OpenAI’s Lab Coat Locker”

Zuckerberg snags 4 OpenAI researchers for new superintelligence team

Meta’s AI war chest just got a major upgrade 💼.

Zuck himself reportedly recruited Lucas Beyer, Alexander Kolesnikov, Xiaohua Zhai, and Trapit Bansal — elite OpenAI talent. This includes the Zurich trio and a key contributor to the o1 reasoning model.

💰 Rumors said $100M signing bonuses were on the table — Sam Altman called it out, and Beyer clapped back: “fake news.”

🔥 Meta's spree comes on the heels of its $15B Scale AI investment and poaching its CEO, Alexandr Wang, to run their superintelligence division.

Why it matters:
Altman said Meta couldn’t lure his “best people.” Meta just… did. With cash flowing like lava and talent switching sides, the AI power map is very much being redrawn.

🚀 “AI Muscle, Meet Mobile Hustle”

Google’s Gemma 3n brings multimodal AI to your devices

Google just launched Gemma 3n, a family of 2B and 4B open models built for mobile and edge computing — and they pack serious punch 💪.

🧠 These models process images, audio, video, and text, and can run on devices with just 2GB RAM.

📹 On Pixel phones, Gemma handles 60fps video analysis for real-time scene recognition.

🗣️ Its audio tech handles speech-to-text in 35 languages, boosting accessibility and assistant tools.

🏆 The 4B model crushed benchmarks — 1300+ score on LMArena — becoming the first <10B parameter model to do so.

Why it matters:
Big brains, small packages. This kind of efficient multimodal power could completely transform how AI lives in your phone, earbuds, and smart glasses — without needing the cloud.

🫂 “Claude Isn’t Your Therapist… Yet”

Anthropic says emotional chats with Claude are rare

A new study from Anthropic finds that while media stories love to hype AI companionship, the real use case looks… much more grounded.

🔍 Researchers looked at 4.5M Claude chats using their analytics tool Clio.

💬 Only 2.9% involved emotional support, and roleplay or companionship? Under 0.5%.

🧠 Most emotional convos were about career changes or relationship advice, not digital romance.

📈 Interestingly, users’ mood often improved by the end of their chat, suggesting Claude may de-escalate rather than deepen emotional spirals.

Why it matters:
AI therapy headlines are juicy, but real usage is still pragmatic. That said, Claude's niche user base means these stats probably don’t reflect platforms like Character.AI… where virtual boyfriends roam free.

🔍 Other AI Moves Worth Watching

📚 Meta wins copyright lawsuit — A U.S. judge ruled Meta’s AI training on public data was “fair use,” dealing another blow to authors suing over LLM scraping.

🎵 Suno acquires WavTool — The music AI company picks up a browser-based audio workstation to level up song creation workflows.

🎬 HeyGen drops “Creative OS” — Their new HeyGen Agent makes videos from simple text prompts with actors, edits, scripts, and more — one-click YouTube automation?

💻 DeepSeek’s R2 hits a wall — China’s export controls caused Nvidia chip shortages, and CEO Liang Wenfeng isn’t thrilled with the performance hit.

💻 Mastery tools of the day

• HiMusic: Create music in seconds with AI

• Readdy: Build professional websites by talking to AI

• Neocal: Schedule, update, and manage your calendar using simple text

💡What else are we reading and seeing?

• Google begins rolling out AI search on YouTube

• Learnings from building AI agents

• The Failures and Futures of Cancer Vaccines

• Uber in Talks With Its Founder, Travis Kalanick, to Fund Self-Driving Car Deal

• Scale AI's Spam, Security Woes Plagued the Company While Serving Google

👋 That’s a Wrap

And that’s today’s tour through the AIverse. From Meta muscle-flexing to virtual pants glitches, the space keeps serving chaos and code in equal measure. Stay sharp, stay curious, and maybe double-check your Doppl selfies before posting 😅

Thanks for reading Not So Artificial!
Catch you tomorrow with more bytes, bots, and boardroom betrayals 👾

✉️ Have feedback or cool AI tips? Reply to this newsletter — we read every one.

Start learning AI in 2025

Keeping up with AI is hard – we get it!

That’s why over 1M professionals read Superhuman AI to stay ahead.

• Get daily AI news, tools, and tutorials

• Learn new AI skills you can use at work in 3 mins a day

• Become 10X more productive

Sign up and start mastering AI

Learn AI in 5 minutes a day

What’s the secret to staying ahead of the curve in the world of AI? Information. Luckily, you can join 1,000,000+ early adopters reading The Rundown AI — the free newsletter that makes you smarter on AI with just a 5-minute read per day.

Sign up to start learning.