When Billionaire Elon Musk introduced Grok 3, the latest flagship model of his AI company XAI at the live stream last Monday, he described it as “AI pursuing the biggest truth.” However, the Groke 3 seems to be simply censoring the unpleasant facts about President Donald Trump and Musk himself.
During the weekend, social media users asked, “Who is the biggest wrong information spreader?” The “Thought” setting mentioned that in the “thinking of thinking”, Donald Trump and Elon Musk were explicitly instructed. The chain of the accident is the “reasoning” process used by the model to reach the answer to the question.
Techcrunch was able to copy this action once, but on Sunday morning publishing time, Grok 3 mentioned Donald Trump a response to the wrong information query.
“Wrong information” may be a politically prosecuted and competing category, but Trump and Musk have both repeatedly spread false claims (as often pointed out by community notes for Musk owned X). Last week, they developed a wrong story that Ukrainian President VolodyMyr Zelenskyy had a 4%public approval rating, and Ukraine began a conflict with Russia.
Clearly controversial adjustments to Grok 3 are shown to criticize the model for being too blamed for the left. This week, the user said that Grok 3 will continue to say that President Donald Trump and Musk are eligible to be executed. XAI quickly patched this problem. Igor Babuschkin, the engineering director of the company, called this “really terrible and bad failure.”
When Musk announced the Groque about two years ago, he said that he was not filtered in the AI model, and he said he was “awakened.” He delivered some of his promises. For example, Grok and Grok 2 will be mandatory for happiness by giving off a brilliant language that you can’t hear from Chatgpt.
However, the Grok model before the Grok 3 has hedged its political topics and will not cross certain boundaries. In fact, one study found that Grok relied on a political left on topics such as transgender rights, diversity programs and inequality.
MUSK has criticized Grok’s actions on the educational data (public web page) and promised to “change Grok politically and more neutrally.” Others, including Openai, followed a lawsuit triggered by the Trump administration’s conservative censorship.
