Welcome, AI Trailblazers!
Meta just hired four top OpenAI researchers, including the entire Zurich trio and a key o1 architect, as it builds an elite AGI unit. Altman downplayed the exits, but the optics say otherwise. Meanwhile, he publicly blasted The New York Times over ChatGPT privacy, Google dropped a virtual try-on app, and a new AI studio wants to mass-produce startups like TikToks.
📌 In today’s Generative AI Newsletter:
Meta hires four from OpenAI, including Zurich leads and o1 brain
Altman slams NYT onstage over data privacy and legal fights
Google’s Doppl app turns your photo into an AI dressing room
Audos studio wants to launch 100K AI startups a year
Special highlight from our network
🔐 𝐖𝐨𝐫𝐥𝐝’𝐬 𝐟𝐢𝐫𝐬𝐭 𝐄𝐧𝐜𝐫𝐲𝐩𝐭𝐞𝐝 𝐆𝐮𝐚𝐫𝐝𝐫𝐚𝐢𝐥𝐬 𝐟𝐨𝐫 𝐎𝐩𝐞𝐫𝐚𝐭𝐢𝐨𝐧𝐚𝐥 𝐀𝐈 𝐒𝐞𝐜𝐮𝐫𝐢𝐭𝐲
Tumeryk + DataKrypto just launched the world’s first Encrypted Guardrails, closing the last mile of AI security. Now, every token - from retrieval to response — is encrypted, policy-aligned, and cryptographically protected.
𝐈𝐟 𝐲𝐨𝐮 𝐚𝐫𝐞 𝐚𝐧 𝐄𝐧𝐭𝐞𝐫𝐩𝐫𝐢𝐬𝐞 𝐨𝐫 𝐚 𝐌𝐚𝐧𝐚𝐠𝐞𝐝 𝐒𝐞𝐫𝐯𝐢𝐜𝐞 𝐏𝐫𝐨𝐯𝐢𝐝𝐞𝐫:
👉 Deploying a Medical Transcription Agent or a Financial Services Chatbot?
👉 Preparing for ISO 42001 to assure your customer’s data is safe?
👉 Concerned about Customer Service AI going rogue & hallucinating?
𝐓𝐡𝐢𝐬 𝐢𝐬 𝐭𝐡𝐞 𝐨𝐩𝐞𝐫𝐚𝐭𝐢𝐨𝐧𝐚𝐥 𝐨𝐛𝐬𝐞𝐫𝐯𝐚𝐛𝐢𝐥𝐢𝐭𝐲 𝐚𝐧𝐝 𝐞𝐧𝐟𝐨𝐫𝐜𝐞𝐦𝐞𝐧𝐭 𝐥𝐚𝐲𝐞𝐫 𝐲𝐨𝐮 𝐜𝐚𝐧’𝐭 𝐚𝐟𝐟𝐨𝐫𝐝 𝐭𝐨 𝐢𝐠𝐧𝐨𝐫𝐞.
𝐄𝐧𝐜𝐫𝐲𝐩𝐭𝐞𝐝 𝐆𝐮𝐚𝐫𝐝𝐫𝐚𝐢𝐥𝐬 𝐬𝐞𝐜𝐮𝐫𝐞 𝐲𝐨𝐮𝐫 𝐀𝐈 𝐬𝐭𝐚𝐜𝐤 𝐞𝐧𝐝 𝐭𝐨 𝐞𝐧𝐝:
☑️ Real-time encryption/decryption of RAG data, model weights, and prompts.
☑️ Prompt inspection only inside secure enclaves.
☑️ AI Trust Score™ quantifies the risk of all prompts and responses.
𝐂𝐮𝐫𝐢𝐨𝐮𝐬 𝐡𝐨𝐰 𝐄𝐧𝐜𝐫𝐲𝐩𝐭𝐞𝐝 𝐆𝐮𝐚𝐫𝐝𝐫𝐚𝐢𝐥𝐬 𝐟𝐢𝐭 𝐢𝐧𝐭𝐨 𝐲𝐨𝐮𝐫 𝐀𝐈 𝐰𝐨𝐫𝐤𝐟𝐥𝐨𝐰?
Special highlight from our network
Building a Lean AI Startup? You’re not thinking big enough.
Henry Shi’s AI Leaderboard has become the map. $1M ARR/employee. Hyper-efficient teams. Insane speed.
But here’s the twist: If you rethink product, you’ve got to rethink distribution too.
That’s where we come in.
We’re bringing Henry Shi — founder of Super.com and curator of the Lean AI Leaderboard — for a rare LIVE session on what building in this new world actually takes.
And we’ll go deep on the other half of the startup equation:
👉 Demand Generation
👉 Audience Engine Design
👉 Distribution-as-Product
If you’re building anything AI-native, this conversation is your north star.
🗓 July 1st, 2025
⏰ 10:30 AM PST
🎣Meta Just Hired Four OpenAI Researchers for Its Superintelligence Dream
Image Credits: Craig T Fruchtman / Getty Images
Meta has poached four top AI researchers from OpenAI, including the entire Zurich founding trio and a key architect of the o1 reasoning model, as Zuckerberg races to build a next-gen superintelligence team with unlimited budget and zero subtlety.
Here’s what went down:
Zurich exodus: Lucas Beyer, Alexander Kolesnikov, and Xiaohua Zhai, who launched OpenAI’s Zurich office, confirmed they’re heading to Meta
o1 brainpower lost: Trapit Bansal, who worked on OpenAI’s high-profile o1 reasoning engine with Ilya Sutskever, is also joining the new Meta unit
$100M offer? “Fake news”: Beyer denied the rumored signing bonuses Sam Altman mentioned last week, calling them “fake news” on X
Zuck’s recruiting machine: Comes after Meta’s $15B Scale AI investment and its controversial hiring of CEO Alexandr Wang to lead this division
Despite Altman’s public claim that “none of OpenAI’s best people” are leaving, Meta’s hiring spree tells another story. With Zuckerberg personally WhatsApping researchers and hosting cozy dinners at his Lake Tahoe estate, the gloves are clearly off and Meta’s superintelligence ambitions just got very real.
🎤 Sam Altman Blasts NYT Over Privacy in Fiery Live Interview
Image Credits: Hard Fork Live / Platformer
OpenAI CEO Sam Altman came out swinging at The New York Times during a live taping of the Hard Fork podcast, using his first minutes onstage to slam the publication’s lawsuit and privacy demands. The unexpected clash gave a rare unfiltered glimpse into OpenAI’s legal and strategic frustrations.
Here’s what went down:
Altman hijacks the mic: Before the interview could even begin, Altman cut in with a jab: “Are you going to talk about where you sue us because you don’t like user privacy?”
Privacy flashpoint: He criticized the NYT for asking OpenAI to retain private ChatGPT logs, calling it “a position we feel strongly about” especially for users who opted to delete their data
COO gets a dig in too: When asked if Zuckerberg actually believes in AGI or is just recruiting, Brad Lightcap quipped, “I think he believes he is superintelligent”
Cracks with Microsoft: Altman acknowledged “flashpoints” in OpenAI’s relationship with its biggest partner, saying both companies are ambitious and sometimes clash
OpenAI may be building the future of intelligence, but right now, its leadership spends just as much energy dodging lawsuits, swatting at rivals, and defending user data. Altman says he still “loves The New York Times,” but onstage in San Francisco, he made it clear: not when they come for OpenAI’s privacy stance.
👗 Google’s New App Lets You Try On Any Outfit With AI
Image Credits: Google
Google just launched Doppl, an experimental app that turns a photo of you into a virtual mannequin so you can preview how any outfit might look on your own body. Available now on iOS and Android in the US, Doppl turns screenshots or street photos of clothing into realistic try-on experiences, complete with AI-generated video.
What’s new:
Upload yourself: Users snap a full-body photo and Doppl builds a 3D version of them to try on outfits
Try anything, from anywhere: Use a thrift store photo, a social media screenshot, or a friend’s look — no purchase needed
Watch it in motion: Doppl generates short clips so you can see how the clothes move on your body
Save and share: Users can bookmark outfits or send their virtual looks to friends
Built on Google Shopping tech: Doppl moves the try-on feature out of search and into a fun standalone app
Unlike Google’s previous try-ons that used preset models, Doppl puts you at the center while giving Google a fresh stream of body data in return. It’s an early Labs experiment, so expect some clunky fits. But the direction is clear: Google wants to be the mirror you scroll through before you dress.
🚀 This AI Startup Studio Wants to Launch 100,000 Companies a Year
Image Credits: Audos
A startup factory called Audos says it can help everyday people launch AI-powered businesses at massive scale. Founded by BarkBox co-creator Henrik Werdelin, Audos uses AI to guide users from idea to launch, aiming to spin up 100,000 small companies a year, each powered by a single founder and social media traction.
What’s new:
No tech skills needed: Founders just chat with an AI agent, which helps shape their idea and test demand using paid Instagram ads
Revenue-share model: Audos takes 15% of future earnings in exchange for tools, up to $25K in funding, and marketing help
Low hundreds live: Early users include an AI golf swing coach, a digital mechanic’s assistant, and “after death logistics” services
Backed by top VCs: Raised $11.5M from True Ventures, Bungalow Capital, and angels like Skype’s Niklas Zennstrom
Werdelin isn’t trying to build the next Stripe. He’s betting on a future where millions of people create $1M businesses using AI instead of VC. Whether that’s empowerment or just a new app store tax depends on how far Audos’s tools can really take its solo founders.
🚀 Boost your business with us. Advertise where 12M+ AI leaders engage
🌟 Sign up for the first AI Hub in the world.
