Grok’s new cursing chatbot follows Elon Musk’s plan to provide an uncensored alternative to ChatGPT. This week, xAI released a new voice interaction mode for its Grok 3 AI model, which is currently available to premium subscribers.
This feature is somewhat similar to OpenAI’s Advanced Voice Mode for ChatGPT, according to Ars Technica. But unlike ChatGPT, Grok offers several uncensored personalities for users. Currently, these are all voiced by the same default female voice. Available modes include a disconnected mode and a version that reenacts verbal sexual scenarios.
Earlier this week, AI researcher Riley Goodside drew attention to the extreme detached mode by tweeting a video in which he repeatedly interrupted the voice chatbot. When he asked the chatbot to shout louder, it started screaming. By default, the disconnected mode uses coarse language, constantly cursing and belittling the user.
Large number of speech personalities
In addition to this extreme mode, Grok offers other speech personalities. These include a Storyteller that tells stories, a Romantic mode that stutters and talks uncertainly, and a Meditation mode that guides users through a meditative experience. There is also a Conspiracy mode that likes to talk about conspiracy theories, UFOs and Bigfoot.
This is in addition to an Unauthorized Therapist who plays the role of a conversational psychologist and a Grok Doc who pretends to be a doctor. Furthermore, the system includes a Sexy mode, marked as 18+, which functions as a phone sex line. A Professor focusing on scientific topics is also available.
This approach by xAI is the opposite of other AI companies, such as OpenAI, which censor conversations about non-safe topics or risky scenarios. For example, Sexy mode will discuss explicit sexual situations, something ChatGPT’s voice mode does not allow. However, OpenAI recently relaxed the moderation of ChatGPT’s text version, which now allows some erotic conversations.
Fewer restrictions than ChatGPT
What we see with Grok’s voice mode seems in line with Elon Musk’s original plans for xAI when he founded the company last year. Musk announced that he wanted xAI’s chatbots to provide uncensored answers, unlike ChatGPT, which he considers too restrictive and politically leftist. Earlier, we saw that xAI allowed Grok to generate largely uncensored images through the X-social networking platform.
Technologically, it is novel to interact with a voice AI chatbot that does not censor itself and is officially provided by a technology company rather than as a jailbreak or open-source hack. This is probably the first chatbot with these capabilities.
Still, tests showed that Grok’s voice mode often repeated itself and stuck in loops, almost as if rehashing pre-programmed talking points. It is not nearly as fluid as ChatGPT’s Advanced Voice Mode, but it is provocative – which is probably exactly what Musk wants to achieve with it.