Grok can now also be marked in posts on X – and replies

Elon Musk's artificial intelligence (AI) called Grok has been available to X users free of charge for some time now –. It now also responds to posts in which it is tagged. It has already answered various posts with questions.

Various users are currently reporting the new option of tagging Grok in their own posts and asking questions, for example, which the AI then answers. As Grok explains when tagged, there is no limit to the number of replies to a post. To chat with Grok in public, it is therefore sufficient to simply reply to the post again without having to tag the AI again.

In a test, however, Grok did not respond to a request to reply again (as of March 7, 5 p.m.):

Grok says it can reply in several languages, including English, Spanish, French, German, Italian, Portuguese, Dutch, Russian, Chinese, Japanese and Korean. The AI draws on its extended multilingual capabilities in the Grok 3 update.

As always, the restrictions are kept to a minimum – When asked whether Grok has certain restrictions in its public responses, the model mentions depictions of violence, offensive content, illegal activities and dangerous behavior.

Most recently, Grok 3 also introduced a voice mode. It can moan, but also scream, swear and say dirty or sexualized sentences. Specifically, the voice mode is only available in a single female voice, but moods and styles can be defined for it, which are not always entirely PG.

And premium users can apparently also use these options in public conversations if they mark Grok with their wishes. According to Grok, the entire range of uncensored language personalities is included, including the "Unhinged" mode and role-playing scenarios for X Premium+ and SuperGrok subscribers.

In Unhinged mode, Grok is expected to respond in a deliberately offensive, inappropriate and abusive manner. According to Grok, compliance with the guidelines is also partially controlled by humans. Unhinged mode may have gone a little beyond the scope.

(nen)