Hacker News Reader

I cannot think of any other reason why the new blog post wouldn't have mentioned the obvious connection to the earlier issues that they had. They want to make it seem like ChatGPT invented this use case but they know that the sample code that ChatGTP learned from was mentioned in their previous blog post.

View on HN

There's a vast chasm between "whoever wrote this article didn't think to link to a similar issue a year ago" and "the first incident was a malicious hoax".

I don't understand the original comment to suggest that. Rather: it's a known issue. ChatGPT does nothing new, and certainly doesn't do it by itself -- it just rehashes what others have already written. Like Google might send you visitors for something that's not even present on your website because others link to you mentioning it.

What the comment suggested was that they're now bringing this up again to get attention (and links) since it's combined with ChatGPT. That's not "malicious", but it's also not exactly "wow, we just realized this happens".

What the comment suggested is that the company deliberately created tools using their own API in a wrong way in order to write a blog post about it.

If that's not an accusation of being malicious I don't know what could be.

Reply on news.ycombinator.com