OpenAI Alleges DeepSeek Used Its Models for AI Training

[MacRumors]

OpenAI says it has uncovered evidence that Chinese AI startup DeepSeek used its proprietary models to train a competing open-source model, potentially violating the company's terms of service.

The discovery centers around a technique called "distillation," where developers use outputs from larger AI models to train smaller ones. The practice is common in AI development, but OpenAI claims DeepSeek crossed a line by using it to build a rival model.

"The issue is when you take it out of the platform and are doing it to create your own model for your own purposes," a source close to OpenAI told the Financial Times.

DeepSeek's R1 reasoning model has attracted widespread attention in the tech industry for achieving comparable results to leading US models at a remarkably low cost. The company claims it spent just $5.6 million on development, which is a fraction of what companies like OpenAI and Google typically invest. The app this week reached the number one position on Apple's App Store free charts in multiple countries, including the US.

Asked about OpenAI's allegations in an interview with Fox News, White House AI czar David Sacks didn't mince his words.

"There's substantial evidence that what DeepSeek did here is they distilled knowledge out of OpenAI models, and I don't think OpenAI is very happy about this," he said.

The controversy has already had market implications. Nvidia saw its shares drop 17% on Monday, wiping a one-day record $589 billion off its market value, as investors questioned whether expensive AI hardware investments might be unnecessary if companies can achieve similar results with fewer resources.

According to Bloomberg, OpenAI and Microsoft reportedly investigated and blocked accounts in August for suspected terms of service violations, and they now believe these accounts were associated with DeepSeek. Both companies have declined to provide specific details about their evidence.


Article Link: OpenAI Alleges DeepSeek Used Its Models for AI Training

 

[madmin]

There's never been much honour between thieves

 

[zilchfox]

OpenAI: “How dare, we trained our AI models on the internet without permission from anyone first!”

Forgive me for playing the world’s smallest violin.

 

[rikscha]

Since when has there been any groundbreaking innovation coming from China? It is always copy and paste, which is seen culturally as some sort of recognition of the great work of the teacher. This concept doesn’t work in a global economy, though, for obvious reasons.

 

[DrJR]

StackOverflow would like to talk to you......

 

[Saturnine]

I was waiting for somebody to propose that this may be the case. I am not surprised by the outcome.

With that said, training an AI model on other information created by other people... seems like par for the course really.

 

[Dr McKay]

After OpenAI trained on all that copyrighted material then gave a half hearted “We’re sorry we got caught, it was totally an error guys!”

Let me grab my smallest violin.

 

[rgeneral]

UFC. AI edition.

 

[10anta]

A cheap counterfeit Chinese copy of a Western product - I’m shocked

 

[DrJR]

Since when has there been any groundbreaking innovation coming from China? It is always copy and paste, which is seen culturally as some sort of recognition of the great work of the teacher. This concept doesn’t work in a global economy, though, for obvious reasons.

 

[johnsawyercjs]

They'll get away with it. It's China, Jake.

 

[cjsuk]

Hey you bastards stole content from my web site and trained your stuff on it. Go screwey!

 

[WarmWinterHat]

They'll get away with it. It's China, Jake.

As they should, OpenAI is getting away with ripping off material from anywhere and everywhere they want, with or without the owners permission.

 

[wanha]

OpenAI should sit this one out, but considering we're neck deep in the age of hypocrisy, I'm not holding my breath on that

 

[[AUT] Thomas]

OpenAI should please enlighten us how they trained their model and how they treat copyrighted material.
A lot of smaller websites will lose visitors because OpenAI and others crawled them and serve the answer directly.

Unfortunately for the AI companies, they are on extremely thin ice and in a glass house. OpenAI, better sit still and ****...

 

[TechnoMonk]

I have no sympathy for closed AI. There is nothing open about it, after getting hundreds of millions in spirit of being open. Deepseek is a game changer.
Open AI: we can train on data from internet, but others can’t.

 

[Chuckeee]

Garbage In, Garbage Out

 

[God of Biscuits]

And heeee-eeere we goooooooo!

They'll all try to discredit the new one.

 

[macduke]

Sucks when somebody takes your hard work without your permission and uses it to train an AI model, doesn't it, Sam?

It sucks when the AI comes for you and takes your job, doesn't it, Sam?

The name OpenAI is a joke. There is nothing open about their approach anymore. It was fine to steal when they were open source, and now that they have a great model they close it down so that they can enrich themselves instead of the world. They got the first mover advantage, they followed the Zuck mantra of "move fast and break things" and now they're upset when someone else does it to them. Give me a friggin break you dweeb. At least the DeepSeek project is open source and can be run on your own hardware!

 

[TechnoMonk]

Since when has there been any groundbreaking innovation coming from China? It is always copy and paste, which is seen culturally as some sort of recognition of the great work of the teacher. This concept doesn’t work in a global economy, though, for obvious reasons.

Actually what deep seek did was far from copy paste. They leveraged RL to cut through labeling and intensive training. In fact, one of they key founding scientist of Open AI called it beginning of a new direction.
Open AI was touting how others can’t train without huge compute advantage and resources. Deep seek showed, you need few million to train instead of billions.

 

[TylerL]

OpenAI is accusing someone of taking their proprietary AI, and turning it into an…open AI.

They’re going to twist themselves into knots explaining why it’s ok for them to scrape everyone’s copyrighted data, but NOT ok for someone to scrape theirs.

 

[Skyscraperfan]

Terms of Service? Open AI used anything it could find for training. So how could they claim any copyright?

 

[johnsawyercjs]

As they should, OpenAI is getting away with ripping off material from anywhere and everywhere they want, with or without the owners permission.

This is true

 

[God of Biscuits]

But Sam Altman, "it's impossible to create DeepSeek without use of copyrighted material from the internet."

https://www.windowscentral.com/software-apps/openai-admits-needs-copyright-materials-for-chatgpthttps://www.windowscentral.com/software-apps/openai-admits-needs-copyright-materials-for-chatgpt

 

[IIGS User]

I downloaded the app and I was using it a bit last night. I went to fiddle with it this morning and it got brain freeze. Couldn’t even answer basic questions.

I smell a lot of hype. The short sellers probably made giant bags of money the past few days. They’ll throw those winnings back into the market and ride the rebound.

I think it’s a flash in the pan, much the same as AI in general. Until it’s smarter than the cat, it’s just a bunch of hype.