In conversations around Open Source AI, a familiar idea keeps surfacing: that “openness is a spectrum.” It sounds convenient, especially when trying to reconcile proprietary data or partially open models with the ideals of Open Source. But let’s be clear: this is a flawed premise—and one the Open Source community must push back on.

Freedom isn’t a sliding scale

The concept of freedom lies at the heart of Open Source. And freedom, by its very nature, is binary. You either have it, or you don’t.

Software that carries restrictions on usage, modification, redistribution, or access to its building blocks (like data or model weights) isn’t partially open. It’s not Open Source at all. Those conditions—ethical restrictions, commercial limitations, time-delayed releases—are forms of restriction, not openness.

The Definition is clear—for a reason

The Open Source Definition exists precisely to make this boundary clear. It doesn’t weigh licenses on a sliding scale of freedom—it simply asks: does the license grant the freedoms to use, study, modify, and share the code?

If it does, it’s Open Source. If is does not, it isn’t.

The same must apply to AI systems. If a model is released without detailing its training data, or under terms that limit how it can be used or shared, it doesn’t matter how “open” it feels—it hasn’t crossed the line into true freedom.

Projects differ—their freedom doesn’t

Open Source isn’t about uniformity. Projects can look wildly different while still being genuinely open. SQLite has no public roadmap and few contributors. Kubernetes is backed by a global foundation with a sophisticated governance model. Both are Open Source.

Some projects embed ethical missions (like the UN’s Sustainable Development Goals), others are commercially driven. These values are layered on top of freedom—not prerequisites for it. As a user or contributor, you are free to value what you want in a project, because Open Source grants you that freedom in the first place.

Learning from the “Lost Decade”

During the rise of cloud and mobile computing, Open Source didn’t fully address the consequences of large-scale commercial adoption. Massive platforms forked Open Source projects and repackaged them into proprietary services with little community contribution. This era—Open Source’s “lost decade”—offers a lesson for AI.

We must not make the same mistake again. If we blur the line between open and closed, we risk enabling centralized control behind a façade of openness.

Upholding Open Source principles

For AI to be truly Open Source, it must uphold the same principles that have defined Open Source software for over two decades. There is no “80% open” or “open enough.” The freedom to use, study, modify, and share isn’t negotiable. It’s either there, or it isn’t.

That clarity is what makes Open Source resilient, inclusive, and powerful.

Source: opensource.org