Put your name on the Open Weights Letter.
Twenty-five organizations — NVIDIA, Meta, Microsoft, IBM, Hugging Face, Mozilla, Mistral, the Linux Foundation, Andreessen Horowitz, Y Combinator and more — just told Washington that America leads in AI by keeping models open: free to download, inspect, modify, and run. NVIDIA's CEO joined X for the first time ever just to share it — 11 million views in hours. The companies have signed. In America's 250th year, citizens sign next.
Sign as a citizen of the open AI economy.
Names, organizations, and roles are published on this page for every visitor to see — that's what a sign-on letter is. Nothing else is collected.
Every name makes it harder to ignore.
What the letter says — without the PDF.
Open source already won this argument once.
In the 1980s, betting on open code sounded reckless. Today it runs most of the internet, the U.S. military's research stack, and every major tech company. The letter's core claim: open-weight models are the sequel — a shared foundation the next generation of American builders can stand on.
Open weights put frontier AI within everyone's reach.
Startups, hospitals, universities, factories, and main-street firms can adapt advanced models instead of training from scratch or paying frontier prices for every task — the right model for the right job at the right cost. That's how AI diffuses into the whole economy, not just a few labs.
Competition keeps the gains broadly shared.
When many organizations can build and deploy, rivalry spreads across models, chips, clouds, and applications. Prices fall, innovation accelerates, and the benefits of AI land across the economy instead of concentrating in a few hands.
Control belongs with the people who build on it.
Organizations investing in AI shouldn't fear vendor lock-in or losing what they've built. Open weights let them keep their data, run models on their own terms, and own the specialized capability they create — the letter calls this American sovereignty and prosperity.
Openness is a safety strategy, not a safety risk.
The letter doesn't dodge the risk: released weights can't be recalled. But it argues the answer is more eyes, not fewer — defenders need capable models too, and a world of a few closed systems means a few single points of failure. Transparency lets thousands of researchers find and fix what a closed lab can't see.
Three moves to keep the ecosystem strong.
01Open the compute gates
Expand access to compute for startups and researchers, so building on advanced AI isn't reserved for the deepest pockets.
02Build shared foundations
Invest in public training assets — datasets, tooling, and evaluation frameworks — the common infrastructure every builder can use.
03Don't regulate openness away
Avoid premature restrictions on open models that choke competition or push innovation offshore — and target actual misappropriation, not standard techniques like distillation.
"…keeping the frontier plural…"— from the coalition letter, July 24, 2026
25 organizations opened the ledger.
Read the original letter: Open Weights and American AI Leadership (PDF)