Okay, so this probably shouldn't be a shock to anyone, given that it is actually one of the highlighting features of this product. But last week, X snuck in a new, opt-in setting under your account options that grants X permission to allow it use your posts and activity in the app to train its Grok AI chatbot.
X has now established all the users to this new default, which the user has no choice about; which declares X can use your posts, and your activity with Grok to train and improve the system. This data may be distributed to xAI, technically a third party, and thereby maybe this usage is not included in the scope that users agree to when creating an X account.

Which is apparently why X has included this more detailed option, which is now enabled by default. You can try it out on your account here.

Now to be clear, X has been totally transparent about the fact that it's training Grok on public X posts, as a way to make a more current and current chatbot than competitors.

X owner Elon Musk has not stopped hyping up Grok's lead over other AI tools both in terms of punctuality and censorship (or lack thereof), which in his opinion at least, made Grok a better model for AI development.
It hasn't always been clear, however, on how it is making use of X posts to build the system.

For example, on the Grok official overview, which X termed "an AI search assistant with a twist of humor and a dash of rebellion," X said that :

Like other LLM's currently Grok-1 was pre-trained by xAI on a range of publicly available text data from the Internet up to Q3 2023 and datasets reviewed and curated by AI Tutors, who are human reviewers. Grok-1 has not been pre-trained on X data (including public X posts).

So the Grok-1 language model wasn't trained on X data, yet nowhere in the overview will you find that X is making public X posts.
X's overview of the process is somewhat opaque in this regard, which again, likely is why it is adding this new toggle, which again it'll need to enable compliance with EU laws surrounding the usage of data, and the ability to opt out for users if they so choose.

It's simply not broadcasting the fact that you can limit your data for xAI should you so desire.

Which makes sense, because given the different controversies with Grok, many users might well opt out, thereby reducing X's training data pool.

For example, Grok's news headlines have proven repeatedly to be incorrect and misleading since the system has been mis-interpreting X interactions on various subjects. And when you add the perspectives of Musk on some of these subjects and his relatively lax stand on moderation, there is a likelihood that Grok will disseminate more mis- and disinformation, particularly just before the U.S. election.
Many users of X would not want that data to be included with that, but exchanging user data with a third party, even one that is a subsidiary of X Corp. (X's parent firm), does seem problematic under current terms of the user agreements.

Yet at the same time, Musk is very keen to keep momentum on the project with a view to making his xAI initiative a much more significant part of his empire. Indeed, Musk has reportedly stated that Tesla must invest $5 billion into the venture if it is ever going to be able to compete with the bigger players.

It needs as many X users as possible to keep this setting switched on, though to ensure it remains its key differentiator.

Hence the quiet roll out, now set to come under more scrutiny.

Overall, it is questionable how this has been activated, but you can now opt out of X using your data if you choose.