News & Events Microsoft probes if DeepSeek-linked group improperly obtained OpenAI data -

[BigFootyNews]

Microsoft and OpenAI are probing if data output from the ChatGPT maker's technology was obtained in an unauthorised manner by a group linked to Chinese artificial intelligence (AI) startup DeepSeek, Bloomberg News reported.


Based on security research findings.

Continue reading...

 

[Ambrose Burnside]

Don't all these AI systems "improperly obtain data" in order to build their models in the first place? Using copyrighted material etc?

 

[Chief]

And now DeepSeek is leaking data:

 

[Gralin]

Welcome to the reality of AI

Sam Altman has been busy telling everyone that they've gotta put up with his AI stealing all their data and trawling their websites for the betterment of humankind

But now that he has competition he's crying about fairness

 

[Gough]

You can't copyright people's lived existence.

 

[Mofra]

China has a history of completely ignoring copyright and IP. Why on Earth would anyone expect them to operate any differently in the AI space?

 

[Moonwatcher]

Don't all these AI systems "improperly obtain data" in order to build their models in the first place? Using copyrighted material etc?

Yes, but they also add value to it.

 

[Moonwatcher]

Obviously it's still dodgy when it's copyright material

 

[Gralin]

Yes, but they also add value to it.

How do they add value to it?

 

[Gralin]

China has a history of completely ignoring copyright and IP. Why on Earth would anyone expect them to operate any differently in the AI space?

Its the usual

Its ok when we do this in America to everyone, but its not OK if anyone does it to us line of whinging

America can throw their weight around with most countries but China don't give a shit

 

[Moonwatcher]

How do they add value to it?

By transforming it into weights of a LLM.

 

[Gralin]

By transforming it into weights of a LLM.

ok
how is that adding value to the creator of the material

as far as I can tell its stealing their IP so another company can profit off it

 

[Moonwatcher]

ok
how is that adding value to the creator of the material

as far as I can tell its stealing their IP so another company can profit off it

It doesn't add value as far as the creator is concerned. But , it is a multibillion dollar industry so clearly value is being added. China stealing the weights aren't adding value.

 

[Gralin]

the shareholders will be really proud of you Moonwatcher

i like how its stealing if a Chinese company does it, but its adding value if an American company does it

 

[Moonwatcher]

the shareholders will be really proud of you Moonwatcher

i like how its stealing if a Chinese company does it, but its adding value if an American company does it

LLMs are a new invention. They are a useful invention. China hasn't invented anything. There is an issue with how LLMs source their training data but it is clearly different to what China is doing.

 

[Gralin]

LLMs are a new invention. They are a useful invention. China hasn't invented anything. There is an issue with how LLMs source their training data but it is clearly different to what China is doing.

Is it?
LLMs are trained on data created by people

they feed it books so it can generate stories and replace authors
they feed it pictures so it can replace artists

its straight up theft designed to change who get paid for the work

so they steal artists work and then generate their own slop, say its their slop they own because their LLM "made it"

but you can't train LLMs off each other

you need real data from people

so once they put everyone out of work, ruin all art, break the internet, stuff up scientific research even worse than it is

have a generation of kids that used chatgpt to pass school and don't know shit

and it all collapses

I'm sure going to be glad a few people got rich while the rest of the world burned for it