Grok's new voice mode can curse, scream, demonstrate explicit behavior
26 Feb 2025

Elon Musk's artificial intelligence company, xAI, has recently launched a new voice interaction mode on the Grok 3 AI model that is available to premium users.

The feature provides a number of uncensored personalities for users to choose from.

The personalities are currently delivered through a default female voice and include an "unhinged" mode and one that simulates verbal sexual scenarios.

'Unhinged' mode: A unique feature of Grok 3
Mode details

The "unhinged" mode of Grok 3 has attracted a lot of attention for its unique capabilities.

As AI researcher Riley Goodside pointed out in a post on X, this mode can even simulate a yell when interrupted repeatedly.

"Grok 3 Voice Mode, following repeated interrupting requests to yell louder, lets out an inhuman 30-second scream, insults me and hangs up," he wrote.

By default, this mode uses vulgar language to curse and belittle the user continuously.

Grok 3 offers a range of uncensored personalities
Personality options

Grok 3 also offers a range of other uncensored personalities such as "Storyteller," "Romantic," "Meditation," "Conspiracy," "Unlicensed Therapist," etc.

Each personality is different in its own way and has a different way of communicating.

For example, the "Romantic" mode speaks slowly and hesitantly while the "Conspiracy" mode loves talking about conspiracy theories.

You can even customize these voice modes as per your liking.

xAI's approach differs from other AI companies
Unique strategy

Unlike other AI firms like OpenAI that censor discussions on not-safe-for-work (NSFW) topics or scenarios deemed too risky, xAI has taken a different approach.

The company allows its chatbots to provide uncensored and "based" answers.

This strategy aligns with Musk's vision, aiming for less restrictive outputs compared to ChatGPT, which he views as too politically left-leaning.