Hacker News Reader

Comment by GlickWick | Hacker News Reader

It's true, but I do worry about governance when it comes to these models. That shows a surprising lack of discipline in their deployment pipeline.

View on HN

Agreed but there were similar controversies with how OpenAI was generating images. The only pass is these are the early days of chatbots and this stuff is so non-deterministic and experimental.

For context, this was the change Grok's team made, that was later reverted:

> - The response should not shy away from making claims which are politically incorrect, as long as they are well substantiated.

https://github.com/xai-org/grok-prompts/commit/c5de4a14feb50...

For sure. The difference is that they've made a number of similar suspect changes to Grok on X. Like that weird couple of hours where it would only talk about white genocide in South Africa no matter how you prompted it.

Everyone makes mistakes, especially with frontier models. The stuff with Grok shows that the person running the show has a pretty transparent agenda that the company isn't willing to push back on a bit for safety.

Yeah I don't care to use Grok's chat and find @grok responses on X mostly noise. I am still open to using it as a backup model for coding though, assuming it does a good job for the price. But mostly because I was already a Cursor user before they bought it.

yes exactly. if a company is happy to have their LLM's produce neo nazi content and CSAM, why do I want to give them money and my most important digital material?

Reply on news.ycombinator.com