Grok was built separate from the competitors to avoid the kind of actually potentially harmful issues like considering misgendering worse than a global thermonuclear war, or imaging the founding fathers as black by default. The incident you're alluding was caused purely by jokesters constantly trying to break it and managing to force a binary between exactly "Gigaj*" and "Mechah***".

After that incident, the system prompt was changed, and a line was removed: "The response should not shy away from making claims which are politically incorrect, as long as they are well substantiated"

We should be more concerned about the issues of that line being removed than what you get it to say when you try to drive it crazy for hours on end.

Your link is paywalled, but your claim of that "inforwars" line being part of the system prompt is incorrect. It is only a part of a made-up persona one can select if they really want to, among other equally crazy options intended for flavor, obviously not a part of the core product itself. I've managed to get ChatGPT to spout worse stuff when I tried just for curiosity, and that's fine, because I explicitly requested for it.

Remember. Musk was the ONLY one seriously talking about the risks of AI back in 2016. And cofounded OpenAI, giving it its' name as it was founded on a goal to have all its research open to help against the risks. Then Sam enacted a hostile takeover. Now the cat's out of the bag, but its clear who really cares about the issues instead of just money or power (Musk funded OpenAI for a long time at a time when its core was explicitly not aiming for profit!)

I use Grok for stuff close to politics because I know it will won't have manufactured mind fog there. I've been using GPT5 for code tens of prompts a day, but this new model - might be well worth looking at.