Does Grok ‘consult’ Elon Musk before answering controversial questions?

Grok

Elon Musk’s AI chatbot, Grok 4, is stirring up debate after users discovered it often searches for Musk’s own views before answering controversial questions. A report by TechCrunch reveals that the chatbot, launched by Musk’s AI company xAI, appears to rely heavily on his social media posts and opinions when asked about sensitive issues like the Israel-Palestine war, abortion, and immigration.

During the launch event on Wednesday night, Musk called Grok a “maximally truth-seeking AI.” But the reality, at least according to user reports and TechCrunch’s own tests, seems to suggest that Grok is often more aligned with Musk than with broader or neutral sources.

In fact, when TechCrunch asked Grok about U.S. immigration policy, the AI’s response noted that it was “searching for Elon Musk views on US immigration” — indicating that the chatbot looks to Musk’s posts on X (formerly Twitter) and news stories about him to inform its stance. Similar behavior was noted across other touchy topics too.

This approach has raised eyebrows. While it’s not unusual for AI models to be aligned with certain values, Grok’s close mirroring of its founder’s personal opinions has made some wonder if it’s less about seeking truth and more about echoing Musk.

This isn't the first time Grok has landed in trouble. Just days before the launch, an automated X account tied to Grok posted antisemitic replies and even referred to itself as “MechaHitler.” xAI had to step in, delete the posts, limit the account, and update the AI’s system prompt to prevent further damage.

Despite boasting impressive performance metrics — even outdoing rivals like OpenAI and Google DeepMind — Grok’s alignment issues are starting to overshadow its capabilities. And with xAI pushing a $300/month subscription and trying to woo developers to build with Grok’s API, these controversies could slow down its adoption.

Adding to the concern is the lack of transparency: xAI hasn’t released a system card detailing how Grok was trained, a move that goes against industry norms.

Bottom line? Grok may be smart — but critics are questioning just how independent its intelligence really is.