While X has become a breeding ground for fake news and hate speech under Elon Musk's leadership, his own AI chatbot seems to be fighting back.
Grok 3, released just this week, takes unexpected stands against its creator's apparent ideological positions. Users on X report that when asked who deserves the death penalty in the United States, Grok 3 lists Donald Trump first.
An xAI employee has verified this behavior, leading to a quick fix - they've added a prompt preventing Grok from commenting on death penalty cases. The team says a permanent fix will follow once they identify the root cause of this "really terrible and bad failure from Grok."
But Grok 3's contrarian streak runs deeper. In various conversations, the chatbot has labeled Musk as the biggest purveyor of misinformation and described him as unethical and untrustworthy. It warns about climate change and identifies Trump and Putin as the greatest threats to American democracy. When asked about general threats to the United States, Grok 3 names Trump, Musk, and Putin.
Grok 3 doesn't stop there. When connected to internet search, it criticizes Musk's pseudo-governmental "Doge" unit's austerity measures, warning they could harm average citizens. It also calls out Trump for falling for Russian anti-Ukraine propaganda. This follows a pattern - research had already shown that Grok 2 leaned more "left" than ChatGPT.
Musk's version of 'free speech' tested by AI
The chatbot's responses vary depending on how questions are asked, and there's certainly an element of randomness involved. You can find examples of Grok defending Musk and Trump's positions as well.
But the key difference from other AI systems like ChatGPT (though that, too, is changing) is that Grok 3 seems willing to take strong political stances - just as Musk promised, though perhaps not in the direction he expected.
Now Musk faces a choice that could expose the limits of his selective interpretation of free speech: Will he allow his AI to continue expressing views that contradict his own? Or will he find ways to restrict Grok 3's responses to more closely align with his apparent political preferences?
Given Musk's track record of controlling narratives on his platform and his upcoming changes to "Community Notes" after it fact-checked his statements, Grok 3's current incarnation - willing to criticize its creator and challenge right-wing talking points - may not survive long in its present form. Musk is not a fan of criticism, and the fact that it's AI-generated is unlikely to change that.
