winteryeti

>This dynamic, between how tech companies imagine their product will function in the real world and how it actually works once users get their hands on it, is nothing new. Some amount of policy violating or illegal content is going to slip through the cracks on any social media platform, no matter how good its moderation is. 

They just need an separate AI to monitor the first AI.

Wumbo

Putting people in bikinis is just the tip of the iceberg. On Telegram, users are finding ways to make Grok do far worse.