Moonshot serves Claude instead of Kimi and collects exchanges for model training

MrBuddyCasino 60 points 66 comments September 11, 2026
twitter.com · View on Hacker News

https://xcancel.com/DavidAgranovich/status/20981685228622154...

Discussion Highlights (12 comments)

ahsillyme

This is weird. I can't think of a better compliment that is simultaneously safer. They'll always trail on data generation then, no? So I can't see the point. Best case for moonshot they get to lie about benchmarks that are gamed regardless, is how I see it.

OutOfHere

There is absolutely nothing ethically wrong with other models using GPT and Claude for training. Heck, GPT-Sol even helped train GPT-Luna. There also is nothing wrong with using customer data for training, since millions if not billions of other users benefit from it. Again, even the big players recognize its relevance and do it.

nacs

Claude/OpenAI etc have taken and continue to take literally all data from the internet, printed books, image, audio, and video humans have created in all of existence without permission to train their models. But when same AI company gets "distilled" or it's own AI-generated content used to train other models, it's suddenly immoral or illegal?

segmondy

So how come super genius Fable and Mythos didn't stop them?

nojito

We also get a glimpse into what those models are being used for. >PLA-affiliated surveillance activity. One user that we assess was likely affiliated with the PLA used what they thought was Moonshot’s Kimi model to load surveillance data from a CCTV archive about a single targeted individual. The user asked Kimi to analyze the CCTV data to understand whether the tracked person was behaving abnormally. The CCTV data included video surveillance from hundreds of cameras in Chengdu, including cameras outside PLA facilities, institutes affiliated with the China Electronics Technology Group Corporation, and a major state-owned enterprise (SOE).

tancop

So we're randomly getting free Claude and helping China beat America at the same time? The only problem with it is they might have worse security than Anthropic and your personal info gets leaked, but I don't think it can happen that easy.

AlanYx

The open question here is how is Anthropic retaining these exchanges? If Moonshot is using the API, normally Anthropic would not retain the exchanges, at least that's the promise. If Anthropic is consistently retaining all exchanges from a class of customers because they're "flagged" but not notifying those customers, how can an average customer trust it won't happen to them? If Moonshot is buying accounts and using those rather than the API, wouldn't they set the "no training on my data" flag in the settings so as to go undetected for longer? If so, we get back to the question of why would Anthropic be retaining the exchanges?

monneyboi

I can't be the only one that is getting tired of this. Framing learning from observation as somehow bad. Something literally everybody is doing, model and human alike. It's the process this whole industry is built on. Pretending that this is bad because the other people are also doing what you have been doing, is hypocritical and childish. Meanwhile I'm sitting here looking at some "Flibbertigittering" spinner like some sort of caveman, unable to steer the model when it misinterprets my ambigious prompt because I'm not allowed to see 80% of the output tokens I'm paying for.

ChrisArchitect

[dupe] Discussion on source: https://news.ycombinator.com/item?id=49647300

jst1fthsdys

Why is this taken at face value when it’s from a company that routinely lies about their model killing all humans someday? Especially when the bad actors just happen to be the US enemies.

kadoban

Can these big companies that want to distill not just get the models? Like how many machines at how many different providers are running Opus 5? Nobody is just throwing the model on a thumb drive and taking it home, or grabbing an old drive that somebody was too lazy to scrub or whatever? That would be vastly more illegal, but would it even matter? What would the practical downside be?

wren6991

K3 is served with full reasoning traces available. Anthropic models aren't. If you were served an Anthropic model instead of K3, it would be blatantly obvious. I have little reason to believe this, and Anthropic have every reason to lie about it.

Semantic search powered by Rivestack pgvector
6,278 stories · 57,251 chunks indexed