You’ve probably noticed the narrative we’re being sold. We are in an AI arms race. If the top startups blink, humanity falls behind. We must accept their secrecy, their closed-door policies, and their massive valuations because the stakes are simply too high.
But let’s look at the reality. The entire foundation of this current AI wave—Google’s famous “Attention Is All You Need” paper—was open. Published. Free for anyone to read and build upon. Now, the very startups that climbed that ladder are kicking it away.
The AI “arms race” isn’t a sprint toward human progress; it’s a closed-door scramble to monetize someone else’s open-source breakthrough.
We are told to cheer on this secrecy because the competition is so fierce. But what are they actually building? As one researcher bluntly put it: if we were racing forward on a cancer-curing AI, fantastic, full steam ahead. But that’s not what we’re racing toward. We aren’t racing toward anything. We’ve been going in circles for years.
The twist here is brutal. We think these companies are hoarding secrets because they’re on the verge of Artificial General Intelligence. They aren’t. Genuine breakthroughs have stalled. Instead of inventing the next transformer architecture, these companies are just tweaking the dials on existing models to squeeze out a few extra percentage points of margin.
The irony stings. The same industry that was built on the ethos of open science is now retreating into a fortress of greed, dressed up as necessity. They claim they have to keep secrets to stay competitive. But hiding your research when you built your company on the open science of others isn’t a competitive strategy—it’s a collective action problem that strangles the very innovation they depend on.
You can’t stand on the shoulders of giants and then kick away the ladder while demanding applause for how tall you’ve become.
Look at the top AI startups today. Many are barely publishing. They want you to believe they’re hiding the keys to the future. But the real secret they’re keeping is that the magic has paused. They are optimizing, not inventing. It is far easier to convince investors you are changing the world if you treat your incremental, profit-driven updates as state secrets.
Stop buying the arms race excuse. The next time a tech CEO claims they can’t share their research because of “competition,” ask them what they’re actually competing for. The answer isn’t the future of humanity. It’s just their next funding round.
Real progress demands daylight. What thrives in the dark is usually just a monopoly in the making.
FAQ
Q: Isn't it normal for companies to protect their intellectual property?
A: Protecting a specific product is standard. But there is a massive difference between protecting a product and hoarding fundamental research when your entire existence is owed to an open-source paper published by Google.
Q: If they stop publishing, does AI innovation just stop?
A: It doesn't stop entirely, but it shifts from fundamental breakthroughs to incremental, profit-driven optimization. The field slows down because the collective sharing of ideas is replaced by isolated, redundant corporate silos.
Q: Maybe they aren't hiding breakthroughs, but failures?
A: Exactly. The secrecy is less about protecting genius and more about hiding the fact that the current AI paradigm has plateaued. If they admitted they were just tweaking existing models, their valuations would collapse.