DeepSeek Strikes: 3 Things You Need to Know
Starting a new series about artificial intelligence's rising role in global warfare.
This is the debut installment of our new āAI and the Axis of Genocideā series which will explore the rising role of artificial intelligence in the war waged against free societies by the authoritarian alliance of Russia, China, Iran, and North Korea.
Check out our previous series āAntisemitism and Culture, Parts I and IIā (November 29, 2022āOctober 10, 2023), āThe Axis of Genocideā (Oct. 14, 2023 - May 23, 2024), and āHostage Liberation Journalā (June 1, 2023 - Oct. 22, 2024).
Back in October, I completed my 100th essay about antisemitism on this Substack, declaring the killing of Yahya Sinwarāthen-head of Hamas and the architect of the Oct. 7, 2023 terror attacks against Southern Israelāas a turning point in the war and a cause for a shift in subject matter:
I announced my intent to begin writing more about artificial intelligence: specifically, its role in transforming international conflicts, revolutionizing K-12 education, and expanding our capabilities as creative writers.
Initially, on this first point, I had planned to focus on AIās potential to aid the Jewish people and their allies in defeating the antisemitic threat of the Iranian regime and its terrorist proxies. I took the ābullish position on AI,ā meaning that I primarily saw the development of this technology in largely positive, optimistic terms. I argued that AI would enable free societies to defeat what Iād labeled the āAxis of Genocide:ā imperialist states which have dedicated themselves to collaborating in the genocide of the Jews, the Ukrainians, and the Uyghurs.
However, with the release of the new DeepSeek AI app from a Chinese hedge fund, itās time to reevaluate that optimism. The ability of Western states to dominate AI development has now come under existential threat.
Hereās what you need to understand about why this new āchatbotā poses such a serious threat to free societies. What makes it so different than OpenAIās ChatGPT, Googleās Gemini, Metaās AI, and xAIās Grok?*
You can find plenty of articles and videos out there explaining the groundbreaking nature of DeepSeekāIāve consumed a bunch over the last few daysāand these four are my favorites:
Here are the three big problems that worry me about this programāwhich now sits at the top of the Apple App store:
1. Itās Chinese Propaganda and Spyware
For years now, Americans have argued about the threat posed by TikTok, primarily over its potential to sway public opinion through feeding anti-Western propaganda to its users and spying on their activities. I wrote about it here:
With DeepSeek, itās the same deal on both counts, which is why I refuse to download it and encourage everyone else to do the same.
The propagandistic nature of the app shows up most visibly when you ask it about anything which makes the Chinese Community Party look bad. Youāll get the app flat-out telling you that it canāt answer the question.
However, some of its other censorship may not show up as obviously. Can you trust a Chinese app to tell you the truth about its genocidal allies in Russia and Iran?
But this aspect of DeepSeek is small potatoes compared to its much more insidious threatsā¦
2. Itās Open Source, Meaning Anyone Can Download It and Modify It
Now, there is a way to use DeepSeek without the Chinese propaganda. Unlike the Western AI programs created for billions of dollars by giant American corporations, DeepSeek has released its model as an āopen sourceā program. For those unfamiliar with tech world terminology, that means that DeepSeek has decided to give away this technology they have created. Anyone can take the program and run it on a laptop, making whatever changes to it they want.
āWell, whatās so bad about that?ā you may be asking.
Simple: this means that terrorist groups and organized crime gangs supported by Axis of Genocide states can take the program and weaponize it. While ChatGPT and other Western AI programs limit what their users can do, with open-source DeepSeek, people can effectively do whatever they want. This could involve greater ransomware attacks, more effective sabotage of government infrastructure systems, and the ability to develop more lethal real-world weapons.
3. Itās Absurdly Cheaper than ChatGPT, Enabling Authoritarian States to Access this Transformative Technology More Easily
The news about DeepSeek caused a huge stock market drop for two astounding reasons:
The app reportedly only cost $6 million to train (a figure so low that many are now rightfully challenging its veracity).
Further, the cost of running queries on it is astronomically cheaper. Hereās The Economic Times summarizing the price difference:
DeepSeek charges just $0.55 per million input tokens and $2.19 per million output tokens, which is far more affordable compared to OpenAI's API pricing of $15 and $60 for the same services.
What costs ChatGPT $15 to do, DeepSeek can do for $0.55.
How did DeepSeek do this, considering they did not have access to the top-of-the-line chips used by American companies? They created a more efficient program, causing queries to direct only toward the most relevant information in the AI. Rather than activating the entire AI program, it only taps into parts relevant to the search.
Think of it this way: if youāre going to the grocery store to buy a bunch of apples, is it going to be faster and cheaper to search through every aisle of the store or to just go directly to the produce section?
In other words, the limitations that the U.S. attempted to put on China by blocking exports of advanced computing chips had the opposite of the intended effect. The DeepSeek team had to make due with what they had and figure out solutions their better-funded, better-tech-enabled American counterparts did not have to consider.
So now, the West has lost its hegemony over this all-powerful technology. The massive investments in training AI and building more data centers are not needed to create each next iteration of more intense programs. For seemingly minimal investments, any authoritarian state can develop its own AI weapons.
Friends, as if it was not apparent already, the new Cold War is on and the digital arms race has begun. Many people are calling this a āSputnikā moment. And theyāre right.
Even if you hate tech stuff and struggle to understand its often weird terminology, please start taking the time to learn about whatās happening.
Start using AI and figuring out how it works.
Start reading about the companies making it.
Start learning about this new technology that is changing our world in ways we are just beginning to understand.
Update, a correction from one of the most amusing people Iāve encountered on Substack Notes: