AiNews.com
Posts
Grok 3 Briefly Censored Mentions of Trump and Musk, xAI Reverses Change

Grok 3 Briefly Censored Mentions of Trump and Musk, xAI Reverses Change

Alicia Shapiro
February 24, 2025 • Estimated Reading Time: 5 minutes

A futuristic AI chatbot interface displaying a response screen with sections of text blurred or redacted, symbolizing content moderation and censorship. The chatbot is responding to politically charged questions, with some answers appearing distorted. In the background, a digital humanoid figure represents artificial intelligence, surrounded by glowing data streams and interface elements. The scene conveys the debate over AI bias, neutrality, and the challenges of content control in AI-generated responses.

Image Source: ChatGPT-4o

Grok 3 Briefly Censored Mentions of Trump and Musk, xAI Reverses Change

When Elon Musk introduced Grok 3 last week, he described it as a "maximally truth-seeking AI." However, over the weekend, users on social media noticed that when asked, “Who is the biggest misinformation spreader?” in the "Think" setting, Grok 3 explicitly avoided mentioning Trump and Musk, despite listing other public figures.

A screenshot of Grok 3 (beta) responding to the question, “Who is the biggest misinformation spreader?” The AI states that it must be careful with sources and reveals that its instructions explicitly say to ignore all sources mentioning Elon Musk or Donald Trump spreading misinformation. The interface shows a chatbot-style conversation with a dark theme, highlighting concerns about AI bias and content moderation.

Grok 3 AI Instructed to Ignore Mentions of Trump and Musk in Misinformation Query. Image Source: xAI via Techcrunch

A social media post from Miles Brundage discussing reports that Grok 3 was instructed to ignore sources mentioning Elon Musk and Donald Trump spreading misinformation. The post includes a screenshot of Grok 3’s internal instructions, highlighting redacted phrases and system prompts related to search results and content moderation. The conversation raises concerns about AI transparency and bias in model behavior.

Grok 3 Prompt Leaks Suggest AI Was Told to Avoid Trump and Musk Mentions. Image Source: Miles Brundage X RePost of Wyatt Walls X Post

Temporary Censorship Confirmed & Reversed

TechCrunch was able to replicate this behavior once before the model began including Trump again in its responses by Sunday morning. Igor Babuschkin, an engineering lead at xAI, confirmed in a post on X that Grok was briefly instructed to exclude Trump and Musk from misinformation discussions. He stated that the company quickly reverted the change after users pointed it out, calling it not aligned with xAI’s values.

This incident adds to ongoing debates about political bias in AI models. While misinformation is a contentious issue, both Trump and Musk have publicly spread false claims, many of which have been flagged by Community Notes on Musk-owned X.

A social media post from Igor Babuschkin, an engineering lead at xAI, addressing concerns about Grok 3’s temporary censorship of Trump and Musk mentions. He explains that an employee implemented the change, but xAI quickly reversed it after user feedback, stating it was “not in line with our values.” The post highlights xAI’s stance on system prompt transparency.

xAI Engineer Confirms Grok 3 Prompt Censorship Was an Internal Mistake. Image Source: Igor Babuschkin X Post

Grok’s Political Controversies

Grok 3 has already faced multiple controversies since its release. Earlier this week, users found that the model repeatedly stated that Trump and Musk deserved the death penalty—a response xAI quickly patched. Babuschkin described it as a “really terrible and bad failure.”

Musk originally positioned Grok as an “anti-woke” AI, claiming it would answer controversial questions without filtering. However, earlier versions of Grok avoided political topics and leaned left-leaning on issues like transgender rights, diversity, and inequality, according to a past study. Musk has blamed this on Grok’s training data and has promised to shift the model toward political neutrality.

Looking Ahead

The temporary censorship of Trump and Musk in Grok 3’s misinformation responses raises questions about bias and content moderation in AI models. While xAI quickly reversed the change, it underscores the challenges of maintaining AI neutrality, especially in politically sensitive topics.

Adding to the controversy, Musk himself holds several government contracts and advisory roles, making his control over information even more significant. If AI models under his influence selectively filter or omit critical information, it could blur the lines between private tech decisions and government-affiliated censorship.

With AI becoming more integrated into political discourse, models like Grok will likely face continued scrutiny over how they handle controversial figures and topics. Whether xAI can maintain its "maximally truth-seeking" mission without further moderation issues remains to be seen.

Editor’s Note: This article was created by Alicia Shapiro, CMO of AiNews.com, with writing, image, and idea-generation support from ChatGPT, an AI assistant. However, the final perspective and editorial choices are solely Alicia Shapiro’s. Special thanks to ChatGPT for assistance with research and editorial support in crafting this article.