Stallman Changed My Mind on Open Source. Now AI’s Closed Gates Are the Same Fight—But Way Bigger
(SeaPRwire) – By: Silas Sterling
The open-source community is seething right now. We’ve seen this movie before—companies locking away knowledge behind proprietary walls—but this time, it’s AI, and the stakes are higher than ever. My office was next to Richard Stallman’s at MIT AI Lab in the 80s. We debated for two years: me pushing proprietary control as the only way to advance tech, him arguing for free access to code. He changed my mind, but today’s AI giants seem to have forgotten that lesson.
Stallman said software is knowledge, not just a commercial asset. He wrote GCC, the compiler that still turns most of the world’s code into machine language. Its success wasn’t just his work—it was thousands of contributors fixing bugs and adding features. GNU/Linux followed, powering most of the internet. These weren’t just tools; they were shared knowledge bases that grew stronger with every contribution. Closed systems can’t match that pace because no single company has all the answers.
Back then, critics claimed open source was insecure. They talked about “security through obscurity”—hiding code to keep it safe. But Stallman was right: transparency lets a global community find and fix bugs fast. Obscurity only hides problems until someone exploits them. Open source won that argument decisively, and the modern world runs on its principles.
Now, AI is going the opposite way. Frontier models—the most advanced ones—are completely closed. Even the models marketed as “open” only release the code to run them. They hide the training data and the code that built the model. So you get a pile of magic numbers you can run but can’t explain. Developers are angry because they can’t audit these models or build on them freely. Companies say AI is too dangerous to open, but that’s the same argument used against open source. Closed models leak, get jailbroken, and concentrate power in a few hands.
User sovereignty is dying here. AI is becoming the library of the future, but if a few companies own all the books, we lose control. A doctor using a closed AI for diagnosis can’t verify its reasoning. An engineer can’t tweak it for their needs. We’re trusting oracles we can’t examine, and that’s a recipe for disaster. We can’t let AI become a closed system—we need to fight for open access, just like Stallman did for software.
Author bio: Silas Sterling, veteran kernel contributor and editor-in-chief of Open Source Security Digest, advocates for community-driven tech innovation.