POPA ReconAnthropic: has never advocated for banning open-weights models, calling those without dangerous capabilities a public good.
Anthropic: has never advocated for banning open-weights models, calling those without dangerous capabilities a public good. This separates Anthropic's actual national security concerns (authoritarian AI dominance) from protectionist restrictions that Amodei argues would not address those risks. Source
The excerpt is quoted from a public Telegram channel, which owns its text.