We’re trading our identity for a bit of fun in an app. And we don’t even realize it. AI takes our face and replaces it with one that’s made to sell. Not to show who we are, but to tell us what to buy.
AI filters are those effects we use in apps to instantly transform our faces: they age us, give us cat ears, turn us into cartoons. Now Snapchat has taken it one step further. They’ve turned them into ad tools. They’re called Sponsored AI Lenses. You look in your phone’s mirror and see a brand on your face, a product printed on your skin, a slogan you never agreed to. The AI decides what to sell—and uses your face to do it. You’re laughing, sharing, having fun. But you’ve become a walking billboard. Info
When a group of minds like this gets together, you don’t need to know what they’re doing. You just need to know they mean business.
We’re no longer talking about companies. We’re talking about power squads moving like blocks. And when they move, they shift everything.
Mira Murati, former CTO of OpenAI and one of the brains behind ChatGPT, DALL·E, and more, has launched a new startup: Thinking Machines Lab.
But here’s the real story: who she brought with her. Radford, McGrew, Schulman. People who don’t talk about AI—they build it. They haven’t launched a product yet, but they’ve already secured billions in funding. It’s like Messi, Ronaldo, and Mbappé forming a new team. Let’s see if they win. Info
Images aren’t enough anymore. AI knows how to tell stories now. It can build scenes with a beginning, a middle, and an end. We’ve moved from generative art to generative direction. And it won’t be long before AI sneaks into the videos we watch every day—without us even noticing.
Until recently, AI-generated videos were short and often incoherent. It could spit out a few seconds, but then it would lose the thread. Scenes felt disconnected. Now NVIDIA and Stanford have cracked it. It’s called Test-Time Training: a method that gives AI visual memory—so it remembers what came before. The result? Videos up to a minute long, with narrative flow like never before. They showed a Tom & Jerry sequence that felt more like it was directed by a filmmaker than generated by code. Info
When a company wants to train 100 million Europeans in AI, the first question should be: who decides what we learn?
Because education is power. And when it’s written by the same people who build the tools, we all end up using AI the way they want—not the way we need.
OpenAI just dropped its EU Economic Blueprint: a €1 billion plan to teach Europeans AI skills by 2030. Sounds like good news.
But here’s the catch: whoever controls the training controls the direction. Teachers don’t just teach—they shape what’s worth knowing. And when that power belongs to a private American company, it’s no longer education. It’s influence. Info
Sometimes, AI gets it right. Not to waste our time, but to save it. Maybe even save lives.
When AI is used for science—not just for selling stuff—it actually makes a difference.
Google Cloud and AI2 (founded by a Microsoft pioneer) just announced a $20 million investment. The goal: use AI to help fight cancer.
It’s called the Cancer AI Alliance—a platform built to accelerate discoveries, improve diagnoses, and cut down on research time. This isn’t a demo. It’s not a flashy launch. It’s infrastructure that could transform medicine. Info
In today’s world, those who open their models are stronger than those who keep them locked up. Because if you’re willing to show what’s inside, it means you’re truly ahead.
That’s what NVIDIA just did. And the message couldn’t be clearer.
Nemotron-Ultra is their latest AI model. It’s massive—253 billion parameters.
To keep it simple: think of it as a machine way more powerful at processing and generating complex content.
And here’s the kicker: it’s open source. Anyone can download it, study it, tweak it. And in benchmarks? It’s outperformed big names like Llama 4 and DeepSeek.
NVIDIA didn’t just say, “We’re the best.” They laid the code on the table to prove it. Info
