The World Model Mystery: Why Big Tech Is Keeping Its AI Plans Hidden

Why the Silence?

When I moderated a panel on world models at the All In conference, I found myself drawn into one of the most enigmatic corners of the AI industry. The two main players in this space — Yann LeCun's AMI Labs and Fei-Fei Li's World Labs — have garnered significant attention and funding. Yet, despite their prominence, they remain frustratingly opaque about what they're actually building.

"We'll talk about it when we're ready to talk about it," said Michael Rabbat, VP of World Models at AMI Labs, when pressed for details about the company's current work.

At face value, this reticence might be understandable. AMI Labs is less than a year old and in its early stages of development. But this pattern of silence isn't limited to just one company — it's a widespread trend across the entire world-modeling sector. Even the companies that supply data for these projects, like Physicl's CEO Alex de Vigan, admit they're in the dark about exactly what their clients are doing with the information they provide.

What Are World Models?

World models represent a powerful idea: creating digital representations of physical reality that can be navigated and manipulated by AI systems. At their core, they're about automating spatial intelligence — essentially giving machines the ability to understand and interact with environments as humans do.

This technology could revolutionize many industries. For instance, in robotics, a world model might enable a robot to carry boxes or navigate complex environments more effectively. In the entertainment industry, these models could help create explorable video game worlds or advanced CGI effects. And in autonomous driving, they offer a framework for better understanding and reacting to real-world scenarios.

The potential applications are vast and lucrative — which makes it all the more curious why companies like AMI Labs and World Labs are keeping their development plans under wraps.

Competitive Dynamics

The secrecy around world models seems to stem from a strategic instinct: protect your competitive edge for as long as possible. In an industry where funding is abundant and competition is fierce, revealing too much too soon could attract rivals before the technology is ready for market.

It's not just about preventing others from copying your approach; it's also about avoiding premature competition in a space that could become crowded quickly. If one company announces a breakthrough in humanoid robotics or advanced video rendering, others might rush to join the race — which is why companies prefer to keep their cards close to their chest until they're certain of their path forward.

This mindset mirrors what some call the "dark forest hypothesis" from science fiction: if you don't know who else is out there in the woods, it's best not to attract attention. In the world of AI, this translates to maintaining strategic silence while building the technology that could one day disrupt entire industries.

Implications for Innovation

The current approach to secrecy in the world model space may be shortsighted. While it provides temporary protection, it also limits collaboration and public understanding of how this emerging tech is developing. When data suppliers, like Physicl, don't know what their contributions are being used for, they can't optimize or tailor their services accordingly.

Furthermore, this lack of transparency can hinder progress in the field. Without open dialogue about development goals and outcomes, it becomes harder to identify best practices or align efforts toward common standards. For researchers, investors, and even end users, transparency would help build trust and ensure that innovations serve broader societal needs.

Looking Ahead

As the world model industry matures, I believe we'll see a shift in approach. Early-stage companies may need to balance strategic secrecy with the necessity of building credibility through demonstration and engagement. As the technology evolves, it will likely become more apparent where specific applications are heading — from autonomous systems to digital twins.

However, for now, the silence continues. We're watching a quiet revolution unfold behind closed doors — one that could redefine how machines understand and interact with the world around us. But until these companies choose to speak up, the real story remains hidden in plain sight.