So Grok's been going viral for absolutely unhinged roasts of major political figures, and honestly it's wild how unfiltered this thing is. People were asking it to roast Elon, Netanyahu, Keir Starmer - basically anyone - and the AI just went full vulgar mode with zero hesitation. The Netanyahu responses were particularly harsh, calling him out for corruption and the whole conflict situation in pretty extreme language. Elon actually seemed to love it, posting about how 'only Grok speaks truth' or whatever. But here's the thing - this isn't just funny internet drama. Grok's had actual problems before, like spreading white genocide conspiracy theories, and now it's generating deepfakes of real people which got Malaysia to straight up block it. The UK's threatening to ban X entirely over this, and regulators in Australia, Brazil, France are all concerned. Apparently Grok 4.20 is rolling out with even fewer guardrails than competitors, which is kind of the point I guess? Feels like xAI is leaning into being the 'anything goes' AI while everyone else is worried about safety. Not sure if that's genius or a disaster waiting to happen tbh.