Posted on 07/28/2025 3:15:50 PM PDT by E. Pluribus Unum
China’s unexpected advances in artificial intelligence in recent months have top U.S. researchers baffled.
DeepSeek, seen as a cost-effective AI breakthrough for China, was a wake-up call for U.S. industry technologists, but many appear to have hit the snooze button. DeepSeek’s rapid development of a model matching American frontier AI companies’ outputs for seemingly lower costs caught many national security officials and researchers off guard this year.
OpenAI’s Katrina Mulligan, who is working to foster closer ties between her company and military and intelligence officials, said DeepSeek stunned government and industry sources alike.
Ms. Mulligan told an AI conference in June that before OpenAI released its first reasoning model, her team talked with senior U.S. officials, including some on the National Security Council and at the Department of Energy.
OpenAI told the officials it believed it was four to six months ahead of its competition. Five months later, DeepSeek burst onto the scene.
“What we didn’t expect is that the first AI lab that caught up with us, that had their, that figured out how to replicate the reasoning breakthrough that we had, was not a U.S. lab; it was a Chinese lab,” she said at the AI+ Expo in Washington. “It was DeepSeek, and that was [a] jarring moment, I think, for us and for our competitors and for the U.S. government.”
Earlier this month, OpenAI CEO Sam Altman indefinitely delayed the launch of a much-anticipated open-weight model, citing the need to “run additional safety tests and review high-risk areas.”
Some observers think a Chinese attempt to steal his thunder helped prompt the delay. Just before Mr. Altman’s announcement, the Chinese startup Moonshot unveiled its open-weight model, Kimi K2, which attracted major interest from AI developers and enthusiasts online. Open-weight large-language models are designed to give people...
(Excerpt) Read more at washingtontimes.com ...
I’m sure they’re reading our mail.
A huge problem for the US is that there is very inadequate coordination and collaboration between civilian and advanced military AI in America whereas there is little to no distinction between the two in China
“OpenAI’s Katrina Mulligan, who is working to foster closer ties between her company and military and intelligence officials, said DeepSeek stunned government and industry sources alike.”
She is full of it. She knows that all the big tech companies who develop AI are also selling the same tech to every country they can including China. Big tech is not loyal to the US, even if they are US owned companies. They absolutely cater to a global market with everything they do.
I don't know why that surprises anyone. China has been stealing Intellectual Property from us for decades. China literally has 4 times as many brains as the U.S., everyone claims that Asians are smarter, yet China can't even invent nor engineer their own things.
Chinese AI may seem intelligent, but after an hour it’s stupid again.
Remarkably ignorant of you.
It is simple: ChiCom researchers either produce or they disappear...
Right, this isn’t the 80s or 90s anymore.
There are export restrictions. We’ve not been exporting AI processors (nVidia) to China. Some of it falls under I.T.A.R. laws. You can’t just export these technologies to whomever.
That said, DeepSeek is a different software approach, they’ve not developed their own chips to run it all on. It’s also open source, so you can run it on the powerful hardware anyway. For once, it gives us the ability to copy+1 on China. The other problem China has, DeepSeek neural training (from our models?) contains all kinds of content that is a nightmare for the CCP (free speech concepts, knowledge of outside cultures, law, religion, etc.). I’m hearing that the CCP want to “shut this thing up” as much as it is an advancement.
Put this altogether and it is a big yawn...in terms of China being ahead of everyone and keeping it that way.
“We’ve not been exporting AI processors (nVidia) to China.”
Yes we are, a whole Truck load went to China just the other day. No way that was not an internal setup blamed as “theft”.
Given your stance please explain how it is that our experts are bewildered by Chinese gains. If those gains were stolen from us, wouldn’t we already know them?
I’m not arguing with you, I’m agreeing with you.
Oh, I get it... ;-D Brain fart.
We know it. We know that they know we know it. Yet it goes on. But what no really talks about is that in many instances, they don't need to steal it. We have Chinese grad students, doing research projects, taking internships, etc.
Notice that we don't get many Chinese students in the liberal arts. They are typical engineering and hard science students.
just another cheap knock off and really, more like BI
Bankrupt Intelligence...
Exactly...it did, which is why it made news, because it shouldn’t have. If this is an ITAR violation then nVidia is going to be in serious trouble. It’s exactly why China would be motivated to steal them, it is illegal to supply them...which is my point and is 100%. This would be a 3rd party sale, nVidia wouldn’t be that stupid - they may still have to face consequences though.
Have you ever been involved in a tech company that has to comply with ITAR? I have. It’s no joke, national security types of consequences. There are several countries you must not export to. The claim that we ‘just sell to whoever’ is 100% wrong.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.