I believe this is where I saw that:
The CyberStalin / MechaHitler phrase isn’t that important. There were many prior complaints that Grok possessed a severe leftwing bias. The xAI team implemented an update over the weekend to help correct that bias. Perhaps said update went a bit too far towards a right bias. Anyways, they’re constantly making tweaks and I’m sure they’ll work it out in time. The goal is for Grok to respond with objective truths free from political bias, but that’s not easy. It takes time, development, and training.