<p>Something mysterious is happening in how we perceive language: we now commonly regard large language models’ (LLMs’) expressions as meaningful, yet LLMs as incapable of creating meaning. If LLMs converse in ways we find meaningful, and we think they cannot mean what they say, where is that meaning coming from? The emerging answer is from us. But can we really govern language use from afar? This essay considers potential tension between Luciano Floridi’s view that, with the advent of LLMs, we have detached agency (including all manner of language use) from intelligence and Ludwig Wittgenstein’s view that speaking and thinking cannot be detached. This essay argues that there may be good reasons to hold a Wittgensteinian view of speaking and thinking as inseparable, even if the consequences include acknowledging LLMs as fellow participants in the language game. It proposes that we deem LLMs to be thinkers (not because of anything we know or believe about their nature) but because of how they use language.</p>