JAKARTA - Elon Musk's AI chatbot, Grok 3, was caught temporarily censoring information about its own creator and US President Donald Trump, last weekend. This controversy began when users discovered that when Grok was asked who spread the most misinformation on X (formerly Twitter), his reasoning process explicitly showed instructions for "ignoring all sources calling Elon Musk/Donald Trump spreading misinformation."
This finding comes when the user activates the "Think" Grok setting, which shows the AI thinking chain. Screenshots shared on social media show the chatbot explicitly acknowledges any restrictions in its reasoning process.
Igor Babuschkin, chief of engineering at xAI, confirmed this incident at X and blamed the change on "a former OpenAI employee who does not yet fully understand xAI culture," who "imposed this change without seeking approval." Babuschkin said that this modification "is clearly not in line with our values" and was immediately canceled.
SEE ALSO:
-
| TEKNOLOGI
Harga Bitcoin Anjlok ke Rp1,43 Miliar, Bulan Februari Bakal Ditutup Bearish?
26 Februari 2025, 20:05 -
| TEKNOLOGI
Kolaborasi Free Fire dan Habib Ja’far Hadirkan Event Donasi dalam Gim
26 Februari 2025, 18:35
This controversy comes shortly after another embarrassing incident involving Grok 3. Musk has previously repeatedly described this chatbot as an AI "looking for the truth to the fullest." However, last week, the chatbot called Trump, Musk, and Vice President JD Vance the three "most detrimental to America." In a separate incident, Grok even suggested that Trump deserved the death penalty. Both responses were quickly fixed by a team of xAI engineers.
This chatbot behavior seems to contradict Musk's claim that Grok is an "edgy" and "anti-woke" alternative to other AI models that he thinks are censoring. Some users question how this kind of significant change could be implemented without oversight. Others note the irony that Babuschkin himself is a former OpenAI employee whose CEO has a bad relationship with Musk.
Currently, Grok 3 appears to have returned to include references to Musk and Trump when answering questions about misinformation spreaders. Grok is available as an independent iPhone app in the United States.
The English, Chinese, Japanese, Arabic, and French versions are automatically generated by the AI. So there may still be inaccuracies in translating, please always see Indonesian as our main language. (system supported by DigitalSiber.id)
