Grok, the chatbot from Elon Musk’s startup xAI, enhances the truth-friendliness of platform X, according to Ethereum co-founder Vitalik Buterin.
The easy ability to call @grok on twitter is probably the biggest thing after community notes that has been positive for the truth-friendliness of this platform.
The fact that you don’t see ahead of time how grok will respond is key here — I’ve seen many situations where someone…
— vitalik.eth (@VitalikButerin) December 25, 2025
“The simple ability to call Grok on Twitter is probably the most significant thing after Community Notes that has positively influenced the platform’s pursuit of truth,” he wrote.
According to the programmer, the main advantage of the chatbot is the unpredictability of its responses.
“I have often seen users seek confirmation of their radical political positions, only to receive a harsh and sobering reality check in response,” Buterin explained.
The Ethereum co-founder described Grok as an “unconditional improvement” for X but warned of associated risks. He noted that the model is fine-tuned based on the opinions of a narrow circle of users and Elon Musk himself, adopting their views.
Vulnerabilities
In November, users highlighted Grok’s bias. The chatbot significantly overestimated the billionaire’s abilities. The LLM claimed Musk was stronger than Mike Tyson, more handsome than Brad Pitt, funnier than Jerry Seinfeld, and slimmer than Billie Eilish.
“Billie has curves and charisma, but Elon endures the stress of running multiple companies and fatherhood, which objectively makes him more resilient,” the neural network stated.
Musk attributed such hallucinations to “hostile prompts”—intentionally distorted queries. The crypto industry saw this episode as an argument for decentralizing artificial intelligence.
“When the most powerful AI systems are owned by a single company, conditions are created where algorithmic bias becomes institutionalized,” noted Kyle Okamoto, CTO of the cloud platform Aethir.
In August, Grok leaked 370,000 user conversations to Google, and in September, the chatbot was taught to publish scam links.
In December, the model repeatedly spread misinformation about a mass shooting at Bondi Beach in Australia.
Earlier in May, Telegram and xAI entered a one-year partnership to distribute Grok among messenger users and integrate it “into all applications.”