Our mission — “vibe-coding interactive worlds” — was never just about code-generation. It’s about using AI to give people programmatic control without needing to code. All these clips are *gameplay videos*; infinitely persistent, created by people with have no domain knowledge.

There’s been amazing progress in generative video AI models but … with almost no impact on gaming. Games require not just painting beautiful scenes, but having an interactive simulation with state, structure, mechanics … and unbounded runtime. @moonlake is delivering that!

https://t.co/XCHX3To5WQ

Prompt: “A vibrant international restaurant” Stumbled upon my text to 3D world generation result from a year ago - still one of my favorites :)

Restyling games is much harder than generating a video. Games can’t unintentionally “reset” at every frame: • Your character should stay the same when the camera moves • Damage, lighting, and objects must persist • A room should look the same when you look away and look back

People are massively under-realizing the potential of "AI generation as programmable game mechanics" — especially with how fast we're progressing towards taming model hallucination. In Reverie, I hack together this simple example: 1. Full moon → environmental modifier → player

sign up to our hackathon here:

I'm still amazed by how the best world simulation evolve in the form of games over time: @StupidFunWill 🫡 SimCity (1989) SimEarth (1999) The Sims (2000) The Spore (2008) ARK Survival (2015) GTA series (2026) Extrapolating how the action space and fidelity have grown, we will

AI for science needs fine-grained evals that are careful about knowledge cutoffs. Come see work by @tianyu_hua @ZiyuX @benklieger @sangttruong @liang_weixin @sunfanyun presented right now!

Thanks Mo, just getting started 🙌

We will enable UGC games at GTA VI level fidelity before November 2026

If you work on real-time models, code generation, RL, or ML infra at Meta and are affected by re-org, apply to work with us We already have something better than Horizon Worlds

Trillions will be poured into training action-conditioned world models in the coming decade. The bottleneck is in *action-conditioned* data, and video game is the where this type of data naturally accrue today. GENIE is already trained on the largest video game dataset that

Let’s go @sharonal_lee

paper here:

Works like this deserve a lot more attention. In embodied AI, the most common argument against physics-based simulation is that deformables are hard/expensive to simulate. Offset Geometric Contact just made it 300x faster. It computes vertex-specific displacement bounds to

Congrats @stash_pomichter!

The future GDP 🇺🇸 contribution of these 50 @extraordinary aliens will be $1 Trillion+ From domestic Rare Earth Metal production, space satellite communication, hurricane disaster defense, hyper-efficient photonics energy generation, DeFi to bank billions, technology to empower

😅 we are hiring, DM open

My brother and I spent our childhoods inventing worlds and quests, dreaming of stepping inside them. Today, I’m launching Moonlake AI with $28M to make those worlds real.