Anthropic provides an LLM service.

Anthropic banned the author for doing nothing wrong, and called him an organisation for some reason.

In this case, all he lost was access to a service which develops a split personality and starts shouting at itself, until it gets banned, rather than completing a task.

Google also provides access to LLMs.

Google could also ban him for doing nothing wrong, and could refer to him as an organisation, in which case he would lose access to services providing him actual value (e-mail, photos, documents, and phone OS.)

Another possibility is there (which was my first reading before I changed my mind and wrote the above):

Google routes through 3rd-party LLMs as part of its service ("link to a google docs form, with a textbox where I tried to convince some Claude C"). The author does nothing wrong, but the Claude C reading his Google Docs form could start shouting at itself until it gets Google banned, at which point Google's services go down, and the author again loses actually valuable services.

Then I’m confused about what is confusing you haha.

The absurd language is meant to highlight the absurdity they feel over the vague terms in their sparse communication with anthropic. It worked for me.

It seems just as likely to me that they're just using their terminology inconsistently as it is that they're using it consistently but with that egregious amount of ambiguity. The only thing that I'm confident about is that they're communicating in a very confusing way, and that doesn't really give me any strong insight into whether they're being consistent but vague or just plain vague.