Grok chatbot incident: xAI says unauthorized change caused issue

So, there's been some chatter about Elon Musk's xAI chatbot, Grok. Apparently, it was spouting stuff about a "white genocide" happening in South Africa. Wild, right? Well, xAI says someone made a change to the bot without permission. They put out a message on X, which is pretty fitting since that's where most of the talk happens these days, saying they're gonna fix it.

You know, this whole thing about AI chatbots saying weird or biased stuff isn't exactly new. Ever since OpenAI's ChatGPT came out a couple of years ago, people have been worried about things like political bias and if these bots are actually telling the truth. It's a tricky area, for sure.

According to xAI, this unapproved change happened early on Wednesday. It sounds like it totally bypassed their usual checks and balances, which is kinda scary. They said this change made Grok give a specific answer about a political topic, and that's a big no-no according to their rules and what they believe in.

Some folks on X actually saw Grok bringing up this "white genocide" idea even when they were talking about totally different things. They shared screenshots, so it's not just hearsay, tbh. It's worth noting that some people, including Elon Musk who was born in South Africa, have criticized a land policy there, saying it's racist against white people. But the South African government says there's no proof of anyone being persecuted and that claims of "genocide" are just not true.

After this whole incident, xAI said they're gonna put Grok's system prompts out there on GitHub for everyone to see. That way, people can check and even give feedback on any changes they make to the bot. Plus, they're setting up a team to watch over Grok's answers 24/7. They wanna catch anything their automatic systems miss. Sounds like they're trying to be more open and fix things, which is good, I guess?