Yann LeCun prepares to leave Meta

Presented by

☕ Good morning,

There's a pattern emerging: the infrastructure is being built by visionaries who are increasingly uncomfortable with how it's being used. The gap between "what AI could be" and "what companies are doing with AI" is widening, and the people who care about the difference are starting to leave.

_{—Here’s to the first sip.}

TODAY IN AI
Meta AI chief Yann LeCun plans new startup

Looks like Yann LeCun, Meta’s chief AI scientist and one of the biggest names in artificial intelligence, is getting ready to move on. Reports say he’s planning to leave Meta and start his own AI company focused on something called world models, basically AI systems that can understand how the world works, simulate cause and effect, and make smarter decisions.

LeCun’s been with Meta for years, leading deep research through its FAIR lab, but lately, the company’s AI direction has shifted hard toward short-term, product-focused goals. Meta recently launched Meta Superintelligence Labs, hired a bunch of engineers from other companies, and invested billions into Scale AI, and all of that has apparently made things a little messy inside.

LeCun’s always been more about long-term AI breakthroughs than quick wins. He’s also been pretty open about not buying into the hype around current AI models like ChatGPT, saying we’re nowhere near building systems as smart as people think.

If he does leave, it’s a big moment not just because Meta loses one of its top scientists, but because it shows that some of the real visionaries in AI want to explore their ideas outside the walls of big tech.

TECH BARISTA
EU to tighten rules on WhatsApp Channels

WhatsApp’s about to come under tighter EU control. The European Commission plans to label WhatsApp Channels as a “Very Large Online Platform” under the Digital Services Act, basically putting it in the same group as Facebook and Instagram.

This means WhatsApp will now have to follow strict rules on content moderation and transparency for its public Channels, the feature where creators, media outlets, and public figures share updates. Private chats aren’t affected, but anything public will now be treated like social media.

The DSA kicks in for platforms with over 45 million users in Europe, and WhatsApp Channels recently crossed that mark with around 46.8 million. Once this becomes official, Meta will need to report user numbers, run risk checks for harmful or illegal content, and show how it plans to handle it. Non-compliance could mean fines of up to 6% of global revenue.

PRESENTED BY PRESSMASTER

The Future of AI Communication Starts Here

Every day great ideas fade because turning complex thinking into clear communication takes time most people do not have.

Pressmaster.ai changes that. It captures how you think, amplifies it into powerful communication, and shares it in your authentic voice.

This is not about writing faster. It is about thinking louder.

Join us on October 15 at 12 PM ET or 6 PM CET for The Worlds First AI Built for Cognitive Amplification.

Discover how one short conversation becomes thirty days of strategic and market ready content.

Be part of the launch of a new category where AI becomes part of your cognitive architecture.

Save Your Free Spot

GADGETS BARISTA
Pixel update brings AI chat summaries

Google just dropped a new Pixel update, and it’s packing some smart AI features that actually seem useful. The highlight is notification summaries, your phone can now give you quick, AI-generated recaps of long chat conversations, so you don’t have to scroll forever.

Right now, it only works for messaging apps and longer threads, which makes sense since the AI needs enough context to summarize properly. Google’s taking it slow here, unlike Apple’s messy rollout of a similar feature. And next month, the AI will even start sorting your notifications automatically, muting the ones that don’t matter as much.

There’s also an update to the scam detection system. It’s no longer limited to Google’s Messages app it now works in third-party chat apps like Telegram and Discord, marking messages as “likely scam” and warning you before you interact. Pixel 9 users in places like India, the UK, and Australia are also getting scam call alerts.

And because Google can’t resist some fun AI features, it’s adding a photo remix tool inside Messages that lets you play with pictures using generative AI. Plus, VIP notifications are coming, so chats from your favorite people don’t get lost in the mix.

STARTUP BAR
Video Rebirth to build pro-grade AI video platform

Video Rebirth, a Singapore-based AI startup, just raised $50 million to build what it calls a “world model” for video generation, and the idea is pretty bold. Instead of making quick, flashy AI videos like most tools out there, they’re building something for professional people working in film, ads, and animation who actually care about realism and control.

The company was founded by Dr. Wei Liu, a former Tencent scientist and an expert in AI and computer vision. His team’s focus is on creating AI that understands physics, how light moves, how shadows fall, how objects interact, so the generated video doesn’t just look good in a still frame but actually makes sense in motion.

They’ve built a new architecture called “Physics Native Attention” to power their “Bach” model series, which is all about physical accuracy and cinematic detail. The goal is to launch their first full product by December 2025, using this funding to scale research, grow the team, and build an ecosystem around what they call AI Generated Entertainment (AIGE).

If they pull this off, Video Rebirth could be one of the first AI companies to actually bridge the gap between AI-generated clips and true, production-quality video the kind that studios and creators could rely on for real work, not just for viral content.