It's also a misunderstanding of what many opponents want. e.g I would love to see high-risk deployments, regardless of underlying model size, regulated thoughtfully. There's been loads of hugely negative consequences of high risk deployments of small models.

→ View original post on X — @jeremyphoward