Hacker News new | past | comments | ask | show | jobs | submit login
[flagged] OpenAI Furious DeepSeek Might Have Stolen All the Data OpenAI Stole from Us (404media.co)
1300 points by latexr 22 hours ago | hide | past | favorite | 13 comments





Comments moved to https://news.ycombinator.com/item?id=42861475, which has the more informative of the two articles that this one was lifted from.

Submitters: "Please submit the original source. If a post reports on something found on another site, submit the latter." - https://news.ycombinator.com/newsguidelines.html

Please especially don't submit knock-off articles that jack up the linkbait and indignation. That's what we're trying to avoid on Hacker News. There are enough places to get that hit elsewhere.


I hate clickbait articles that try to make the bad guys seem like they're angry.

> Both Bloomberg and the Financial Times are reporting that Microsoft and OpenAI have been probing whether DeepSeek improperly trained the R1 model

The company openai is not angry, or furious, or enraged. They simply suspect that deepseek broke their usage agreement and are trying to verify that.


I came here to say just this.

Can we change the headline of this article to something more accurate and less clickbaity?

The article unjustifiably labels OpenAI as "furious" despite surfacing zero evidence that that's how they actually feel, obviously in an attempt to paint them as hypocrites who are okay with copying others but are upset at being copied.

This is a very dishonestly-framed and -advertised story.


As I understand from Twitter, the issue explained in this article is not the actual case at hand. The issue is that they suspect them of stealing the o1 model with the weights via corporate espionage and optimizing it with Matrix Multiplication and other upgrades. That would explain why the outputs are nearly identical in some cases.

I don't know how much of any of this is true. This is what I'm reading on Twitter today.


Oh geez yeah, if they really managed to optimize it with Matrix Multiplication these guys must be good

OpenAI's model is closed source. IDK if distilling can be done via the API effectively? DeepSeek already has distilled models from other open source models like Qwen which have been done by 3rd party researchers, and I assumed that happened rapidly because they are all open source.

There is nothing in this article to suggest that OpenAI is "furious" or even upset. Zero evidence. It's total clickbait.

And it's embarrassing that so many commenters on Hacker News who want to believe this storyline are just pretending that it's true despite the lack of evidence.


Such a perfect article title, but wasted on clickbait.

Agree, on reading the title my immediate thought was it belonged to The Register, but would have been even better as the headline of a satirical takedown from Private Eye or The Onion.

It's funny though. There seem to be a lot of commenters on Hacker News who don't really get the sense of humor thing.


[flagged] of course. This place is starting to drift in one direction.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: