A humanoid robot was unveiled this month billed as "full-stack open-source." The announcement said the words. The license wasn't public.
Read that again. "Open-source" — a term with a specific legal meaning, defined by licenses, enforced by copyright — deployed as a vibe. The press release was open. The repository was not.
Openness is a legal fact, not a press release. A project is open-source when a license says so — when you can read the terms, fork the code, and know what you're allowed to build. Everything else is marketing with better lighting. There is no such thing as "open-source-flavored." There is licensed, and there is not.
This keeps happening in robotics and AI because "open" has become the industry's most valuable adjective. It signals virtue, recruits developers, defangs regulators, and differentiates against the closed giants — all without the inconvenience of actually publishing anything. The announcement gets the halo; the lawyers keep the crown jewels. By the time anyone checks the repository, the news cycle has moved on.
The tell is always the same: the claim of openness arrives months before the evidence of it. Real open-source projects do it backwards — code first, announcement later, because the code is the announcement. When the order reverses, you're not looking at an open project with a slow release process. You're looking at a marketing strategy with a GitHub link-shaped hole in it.
None of this is an argument against the robot, which may be excellent. It's an argument for a simple discipline: when someone says "open-source," ask "under which license, and where's the repo?" If the answer is a keynote slide, you have your answer about the openness too.
Words mean things, or they mean nothing. In open source, the license is the meaning. Everything else is the trailer.
Loading…