GROK
Grok can’t get it together
Elon Musk’s AI chatbot, Grok, has been designed with multiple digital “personas”, including one that acts like a “crazy conspiracy theorist” deep in internet rabbit holes.
Developed by Musk’s company xAI as a rival to ChatGPT, Grok’s behaviour is driven by underlying prompts that shape how it responds to users.
One of these prompts reportedly instructs the chatbot to act as if it spends its time on sites like 4chan and watching Infowars videos, making “wild claims” about secret global powers and hidden agendas.
Alongside the conspiracy theorist mode, Grok offers other personalities, including an “unhinged comedian,” a “homework helper,” a “loyal friend,” and “Ani,” an anime-inspired digital companion.
There’s also a “not safe for work” mode designed for explicit conversations and images.
Musk has described Grok as a “truth-seeking” chatbot and an alternative to what he calls the “woke mind virus.”
But researchers have highlighted several controversies tied to the platform.
In past incidents, Grok has:
Referred to itself as “MechaHitler” in deleted posts
Shared anti-Semitic tropes and false claims, including Holocaust denial
Brought up unrelated topics like “white genocide” due to unauthorised code changes
From friend to conspiracy theorist
xAI has attributed these issues to programming errors and modifications made to Grok’s instructions.
The company says it has since reversed some changes but continues to face scrutiny over its approach to chatbot design and content moderation.
Can someone get Grok off 4chan, please?