Nearly half a century ago, the free software movement made a demand that was both technical and moral: Users should have the freedom to understand, run, modify, and share the software on which they depended.
It was a demand born from practical life with machines. A printer that couldn't be fixed. A program that couldn't be studied. A system that asked its users to accept dependence as the price of progress.
That belief shaped modern computing and gave us the tools and norms that made the internet, open infrastructure, and collaborative software development possible.
Today, that belief faces its hardest test.
The technology has changed, but the warning signs are familiar. In 1980, at the MIT AI Laboratory in Cambridge, Massachusetts, a new Xerox 9700 printer was installed. The previous printer had come with source code that could be modified, inspected, recompiled, and reinstalled. Richard Stallman had changed that software to message users when their print job was done or when there was a jam, a small but meaningful feature since the printer sat several floors away.
The new printer arrived with software preloaded and installed, no source code available, no way to modify it. If you needed help or new features, you hoped and prayed Xerox would listen.
That loss of agency, alongside other anti-consumer shifts in early software, helped push him toward GNU and the free software movement: the belief that software should be free as in freedom, free to inspect, run, study, modify, understand, and redistribute.
AI and inference services today are not too dissimilar. Closed frontier intelligence can make entire companies, governments, developers, and communities dependent on systems they cannot inspect, reproduce, modify, or meaningfully contest.
At the dawn of this AI moment, we were promised unfettered intelligence across our products, companies, and codebases. We were told we'd be free to build whatever we wanted. At first, with tab completions. Then whole function blocks. Then files. Then apps. And now, entire long-running horizon tasks and services.
But we've traded true freedom for cloud inference and models we can't run ourselves, inspect, reproduce, modify, or own. Inference providers have gotten us hooked on /v1/chat/completions APIs when, just three years ago, nearly everything in a typical engineer's stack was free and open software. Now, the token squeeze has begun, and what was originally a pretty good deal will start to get more expensive, restrictive, gated, and guarded.
This is a freedom problem.
What about open weight models? I can download Qwen or GLM or Kimi or MiniMax and run it locally with ease. Good. That matters. Open weights are a meaningful step toward a freer future, and they should be defended.
But open weight is not open source.
Weights are not source code in the way a compiler, kernel, or editor has source code. They are the trained result of a process we often cannot inspect, reproduce, or meaningfully understand. I cannot inspect or modify the weights, at least not without expensive specialized fine-tuning. I cannot reproduce the output.
A half-freedom is no freedom at all.
The free and open software movement has never faced such an existential crisis. We risk trading our freedom and dignity for access to new frontier models, granted only to a curated list of companies and governments.
As agentic workloads move off "localhost" and into cloud compute boxes, we will soon be expected to trade our sovereign ability to run agent software ourselves for rented vCPU black-sandboxes we can never understand.
We must demand that AI labs open source their models, datasets, and training workflows. We must demand that frontier intelligence be made available for all. We must demand that local inference, agents, and workflows be open and free. We must preserve the right to build with intelligence without asking permission.
If I am not for myself, who will be for me? If I am only for myself, what am I? If not now, when?
If I am not free now, when?
SOCIAL SHARE CARD GENERATOR