Nvidia and Allies Answer Open-Model Criticism With a New Alliance

After the Hugging Face attack undercut their open letter against restricting open-weights models, many of the same signatories, led by Nvidia, launched an Open Secure AI Alliance that repeats the argument without offering funding or support to vulnerable institutions.

An open letter opposing restrictions on open-weights models drew a poor reception once readers connected it to the Hugging Face attack.
The letter's backers praised a permissive Chinese model used in the response to the attack, framing the absence of guardrails as an advantage.
Critics point out that the same unrestricted capabilities used against Hugging Face would be available to anyone who registers for an account.
Nvidia and other signatories have now announced an Open Secure AI Alliance, which restates the case that unrestricted AI in good hands counters unrestricted AI in bad hands.
The announcement includes no money or technical help for soft targets such as hospitals and schools, only pledges of more open models, weights, data and agent harness research.