Most of us use AI like a rental car service. We hop into a tool like ChatGPT or Claude, ask it to do something, and trust the company behind it to handle everything. But a growing movement of technologists thinks this setup is a mistake that could eventually limit what you are allowed to do and create.
The current race in AI is focused on massive, highly controlled systems often called frontier models. These are incredibly expensive and complex brains built by a handful of large companies. Industry observers, including long-time tech expert Tim O'Reilly, argue that these companies are building a structure of control. They want to move toward open AI, which goes beyond just releasing the underlying math. It means creating a future where you can customize the AI, keep your personal information off their servers, and switch between different tools without losing your history or the way the system has learned to help you.
The difference between renting and owning AI
Think of modern AI as a massive, rigid machine. A model is the core intelligence—the vast collection of internal settings that determine how the AI reacts to your requests. Today, companies keep these settings locked inside their own private silos. They also control the harness, which acts as the digital pipeline or interface connecting that core brain to your specific data and tasks. By controlling this pipeline, companies act as the sole gatekeeper. An open approach would separate these pieces. Imagine if you could swap out the engine of your car while keeping the same steering wheel and navigation system. That is the goal of open-source AI: making the core intelligence something you can host, modify, and integrate into your own life or work, rather than being forced to use only what a giant corporation provides.
The danger, according to critics, is that we are building an internet that looks like the old world of mainframes—where only a few powerful organizations had the keys to the kingdom. If we rely entirely on these closed systems, we lose the ability to innovate on our own terms. More importantly, it creates a privacy risk: if your AI lives entirely on someone else's server, that company has a permanent record of how you think, work, and create. By pushing for open alternatives, the goal is to ensure that AI remains a flexible, personal tool that works for you, rather than a corporate product designed to keep you inside their walls. The future of the technology might not be in the biggest, most expensive model, but in the ones that give you the freedom to build your own path.
Liked this one? The next lands at breakfast.
Every story in tomorrow's AI news, rebuilt in plain English — five minutes, sources linked, free forever.
By joining you agree to receive Article's daily newsletter — unsubscribe in one click. Privacy