It's quite a slippery slope to accept a government's control over speech, even the speech emitted by a machine, because how different really is the text you read from a book, or the text you read from an AI model?
Today for the first time I'm aware of, a government made an AI company turn of it's model to the general public. This is interesting, because all the model does is generate "speech." You could say it emits knowledge. Humans also emit knowledge, and speech. Would it make sense then for the government to shut up humans that could have said or transmitted the same information the AI model could do? That would be an obvious violation of several rights guaranteed under the constitution.
The most slippery of slopes.
re: "Anthropic"