No one is demanding every piece of software be open source. But high-quality open weights should be allowed to exist. Why the resistance? Because strong open models threaten the core of their valuations. When the pure model layer gets commoditized, the trillion-dollar premiums shrink. Open models drive the cost of intelligence toward zero. Do we want the foundational layer of AI stays locked up by a few companies, or do we let the cost of intelligence keep falling so value can flow to the businesses that actually put it to work? It's an obvious answer. It's also so ironic that Anthropic trained on basically everyone’s data and IP whether the creators wanted it or not. Yet they’re pushing rul
No one is demanding every piece of software be open source. But high-quality open weights should be allowed to exist. Why the resistance? Because strong open models threaten the core of their valuations. When the pure model layer gets commoditized, the trillion-dollar premiums shrink. Open models drive the cost of intelligence toward zero.
Do we want the foundational layer of AI stays locked up by a few companies, or do we let the cost of intelligence keep falling so value can flow to the businesses that actually put it to work? It's an obvious answer.
It's also so ironic that Anthropic trained on basically everyone’s data and IP whether the creators wanted it or not. Yet they’re pushing rules that would stop others from using their model outputs to train new models. We can use all of yours! You can’t use any of ours!
