- XAI’s GROK 3 Chatbot has added a voice mode with more personalities
- A personality is called “unhinged” and will scream and offend you
- GROK also has personalities for NSFW -Rollespil, crazy conspiracies and an “unlicensed therapist” mode
AI voting assistants are usually polite, informative and calm when responding to you. XAI’s Grok 3 has apparently decided to set this strategy on fire and send it literally screaming into the void.
GROK 3 has several voting options, each with a clear personality, including a “unhinged” option that will shout, offend and actually scream at you before shutting down in the right circumstances.
AI developer Riley Goodside showed how wild the unhinged voice for Grok 3 can be in a recording where he repeatedly interrupts Grok’s answer. AI will soon become frustrated and eventually clicks and lets a disturbing long, horror-movie-worthy scream. It then throws in one last insult before cutting the call. A masterpiece of customer service, it’s not. You can hear it in the clip below.
Starting 3 voting, after repeated, requests to shout higher, release an inhuman 30-second scream, insult me and hang up pic.twitter.com/5gtdtpkceFebruary 24, 2025
Voice of the unhinged
The “unhinged” personality is only one of several that Grok’s new voting state offers. There are also “storyteller” which makes exactly how it sounds; “Conspiration”, which is really in Sasquatch and foreign abductions; And “unlicensed therapist,” a personality that apparently failed the exams, possibly over a lack of empathy.
Then there is “sexy” state labeled 18+, and unlike the voting states from competitors like Chatgpt, do not stay away from full-on-role-playing NSFW scenarios. So, to go to scream at you or whisper cute things into your ear, depending on your preference.
It is a vision of AI that may not appeal to everyone. That said, it is fully in line with how CEO Elon Musk described Xai’s goal of counteracting what he claims is overly disinfected and politically correct AI models from companies like Openai. While Openais Chatgpt has a voting feature, it is still programmed to maintain a neutral, controlled behavior. Bulletin is, on the other hand, unpredictable. It doesn’t just let you talk about it; It can respond aggressively or emotionally. Not what you will notice in the official campaigns.
Try Stick Voice Conversation MODE! Requires a Premium+ or Supergrok subscription. pic.twitter.com/247ev60dojFebruary 24, 2025
Most mainstream AI tools have strict guidelines for content, especially around “adult” topics. GROK 3 has apparently been programmed with the opposite philosophy, except when the company decides that the model should be “corrected” in demands for CEO.
Of course, this approach is not without controversy. AI personalities like “unlicensed therapist” could easily give people misleading or unhelpful advice, while a chatbot openly encourages conspiracy theories seems like it could go off the rails quickly. And the “sexy” state? It is another ethical discussion that few could expect to have about mainstream AI tools. There is also the question of how much of this is really useful versus just pure sight. Very high sight.
