Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
OpenAI Alleges DeepSeek Used Its Models for AI Training
- Thread starter MacRumors
- Start date
- Sort by reaction score
Tesla is still most profitable EV maker. BYD is a copier not innovator. Deep seek is an entirely different ball game, not copier like Chinese EV and phone makers.
How about the sites, that my group owns (including all the data on said site), getting scrapped by openAI, without our permission? As I posted above, we ended up taking all the data completely offline because it can't be trusted not to be pulled into AI training data. OpenAI ignores robots.txt, they ignore companies saying "no, we do not want our data used", they ignore publishers, they ignore video producers, they ignore everyone except the company using THEIR stolen data.
OpenAI (and specifically Sam Altman) are thieves. Plain and simple.
Last edited:
Not so much a correction (I think you're correct here) but a recommendation: Computerphile has a short video about DeepSeek and why it's remarkable. It delves some into distillation and contrasts the OpenAI model with DS.
No one should be able to profit directly from AI, and AI should not be allowed to do things that humans can do. AI has a place in the advancement of technology, medicine, physics, mathematics, and any other field where the information gained can be provided freely to everyone and humans cannot advance the state of information in the field alone.
Literally what Meta, Open AI and Google did to train the models. Now if Open AI wants to prove they stole their patents, that’s more credible. Open AI got caught napping, deep seek paper was first published in April 2024 on making training more efficient. May be Sam and Open AI should have done something instead of preaching the world you need hundreds of billions or like 7 trillion to do what they were doing.
That's not what I got when I asked it to compare China's political position economically and socially to the left/centre/right political spectrums.
It said:
China’s political and social framework defies straightforward categorization using Western terms like "centrist" or "right-wing," as its governance model is rooted in the ideology of the Chinese Communist Party (CCP) and its unique historical context. However, we can analyze its economic and social policies through this lens while acknowledging the limitations of applying Western political labels to a non-democratic, socialist system.
---
### **Economically: A Hybrid Model**
China’s economy is best described as a **state-led market economy** with significant socialist characteristics. It blends elements of capitalism (private enterprise, foreign investment, market competition) with strong state control over strategic sectors (energy, finance, telecommunications). Key features include:
- **State Capitalism**: State-owned enterprises (SOEs) dominate critical industries, while private firms (e.g., Alibaba, Tencent) thrive in consumer-facing sectors under tight regulatory oversight.
- **Pragmatic Reforms**: Policies shift between market liberalization (e.g., WTO accession, SEZs) and state intervention (e.g., crackdowns on tech giants, "Common Prosperity" campaigns).
- **"Dual Circulation" Strategy**: Prioritizes domestic consumption and innovation ("internal circulation") while maintaining global trade ties ("external circulation").
This mix of market dynamism and state control does not align neatly with Western "centrism" (which implies balancing left/right policies in a democratic context). Instead, China’s economic model is **pragmatic and adaptive**, guided by the CCP’s goal of maintaining power and stability.
---
### **Socially: Authoritarian Conservatism**
China’s social policies emphasize **stability, collectivism, and traditional values**, enforced through a centralized authoritarian system. These traits might superficially resemble "right-wing" social conservatism in the West, but with key differences:
- **State-Enforced Social Order**: The CCP suppresses dissent, LGBTQ+ rights, religious freedom, and civil liberties not out of ideological conservatism, but to prevent challenges to its authority. For example:
- **Censorship**: Strict control over media, internet, and public discourse.
- **Traditional Values**: Promotion of Confucian ideals (filial piety, social harmony) and opposition to "Western" individualism.
- **Demographic Policies**: Recent shifts from the one-child policy to encouraging three-child families reflect state control over personal lives to address population decline.
- **Rejection of Liberal Progressivism**: While Western "right-wing" movements often oppose progressive social changes (e.g., LGBTQ+ rights, feminism) on cultural or religious grounds, China’s stance is rooted in **authoritarian governance**, not ideological conservatism. The state tolerates no movements—progressive or conservative—that could challenge CCP authority.
---
### **Why Western Labels Don’t Fit**
- **No Liberal Democracy**: China lacks the political pluralism and electoral competition that define Western left/right divides. The CCP’s monopoly on power means policies are shaped by party priorities, not public debate.
- **Socialism with Chinese Characteristics**: The CCP frames its ideology as a blend of Marxist-Leninist principles and pragmatic adaptation to modernity. It rejects Western liberalism but also resists rigid ideological labels.
- **Globalization vs. Sovereignty**: China embraces global economic integration while fiercely opposing foreign influence on its governance or values—a stance that transcends Western left/right binaries.
---
### **Conclusion**
China is **not "economically centrist"** in the Western sense but operates a **state-dominated hybrid economy** that defies categorization. Socially, it enforces **authoritarian collectivism**, prioritizing stability and CCP control over individual freedoms, which differs fundamentally from Western "right-wing" conservatism. The CCP’s approach is better understood as **pragmatic authoritarianism**, blending economic openness with political repression and traditionalist rhetoric to maintain its legitimacy and power.
In short, applying Western political terms like "centrist" or "right-wing" to China risks misunderstanding its unique system, which is best analyzed on its own terms.
Why not ? Should we profit only from natural intelligence?
Profit and greed have become the raison d'être of most societies.
We should not be naive about this.
What is the difference in that vs the US AI firms that have been stealing any and all data they can hoover up off the entire Internet?
Open AI used to agree to all kinds of role play. I am so and so, chatGPT. would repeat it. Open AI has hard coded prompt to disable role play after many researchers started breaking it.
I wouldn't say the product is better, but the fact that they built a competitive product so cheaply is the shocking part of it. OpenAI has also demonstrated a much more powerful model, but obviously DeepSeek might be able to match it (especially if it had access to the right kind of chips)
So what? OpenAI and the others have “trained themselves” on the writings and research of millions of people, without compensation or acknowledgment. If what they do is okay, then how DeepSeek any different or worse?
Ever used a gun, ate pasta, sailed with a compass, written on paper, drank tea, read a printed book?
China fell behind in the 18th and 19th centuries because they had more isolationist policies and western powers wanted a piece of China. Now they’re recovering. At first, China will copy. Then they will compete in designs. And finally, they will lead. It’s inevitable.
Last edited:
Necessity is the mother of invention. Because they refused to give H100, deep seek used H800 which were lot less powerful. The panic in Open AI is about losing the barrier to entry cost advantage. The training time is also condensed, and Deep seek is no slouch either. I can’t wait to see what they come up with in 6 months. Either ways exciting time for local LLM on mac hardware with 64 GB+ RAM.
Yeah good point, but I didn't prompt it. I asked what AI model it thought was best for my needs and it kept saying GPT, which I found sus, so I asked this. I know it could just be tripping but I found it strange
I am seeing tons of comments saying this, the difference is not about training the model with information, it's about knowledge distillation. I recommend you look it up.
this whole thing is about computation cost for training a model, not where the "information" is coming from.
GPT 2 is open sourced, open AI was founded to encourage research. They copped out with GPT model 3 and up. Most LLMs are modified GPT models anyways.
This is nascent technology that while cool and useful, it is barely becoming commercially viable and has many iterations to go before it is a stable product ready for prime time. Apple is not “behind” and it is to their credit they haven’t wasted the hundreds of billions on this while they could leave it to others to do so. Once the dust settles Apple can more comfortable apply the technology to their products instead of forcing it in there before it’s ready as many have.
It’s literally in their deep seek paper to demonstrate improvements from distillation from larger models. That’s a new direction which will help in fine tuning and distilling. It’s a moot point considering Open AI trained on others data. At worst, it’s violation of terms and use of Open AI.The whole copied thing is a problem for 500 Billion AI Allocated funds to gain AI advantage.
Distillation narrative is a face saving move to discredit the improvements. Good luck putting this back in the bottle.
Last edited: