High-Flier, a Chinese company founded in late 2016, has recently shaken the American tech and artificial intelligence scene. The company has made an AI software that has been touted as being a competitor to OpenAi’s Chat GPT, which revolutionized the AI scene late 2022. The company released its new model of DeepSeek AI this year on January 10, 2025. The app has immediately received attention from many mainstream media companies and consumers. This comes from fear of a sort of second cold war with China involving AI. The release of the AI effected U.S. stocks almost immediately. Companies like Nvidia and Nasdaq have dropped steeply, being AI tech leaders. Nasdaq has dropped 3.1% and Nvidia has dropped around 17%, losing about 600 billion dollars in market value. The reaction comes from the reported resources that went into making DeepSeek AI. The company behind DeepSeek, High-Flyer, has reported only spending around 5 million in development, which can be assumed already beats the costs put forward to fund Chat GPT. 

However, there are still important differences between DeepSeek and its rival. When put to the test, both models were asked to “write a story about an Indian music artist”. They instantly produced stories- both with similar format, which is expected from AI, as they are simply trained off of existing work, most times current AI models have a dry or cliche writing style. Despite this, I found Chat GPT to be much more convincing and detailed than DeepSeek, though the verdict is quite narrow as the stories were quite similar.  The stories were not exact in words but the texture and “content” of the stories were nearly identical. 

Furthermore, I also put the models head-to-head in a test of Arithmetic. When given a function and four different letter options and asking which of the given answers were correct, Chat GPT answered correctly near-instantly, but DeepSeek provided an answer of 96 when the correct answer was 16. This shows that DeepSeek may not be very reliable when it comes to arithmetic, though it could solve simpler problems such as algebra and finding the circumference of a circle. All around the new DeepSeek model excels at most math problems compared to Chat GPT, but still needs refinement. The model also seems very censored. When I asked it about China related topics, it thought for a while and provided some responses before going back and saying “Sorry, that’s beyond my current scope. Let’s talk about something else.” Many of the topics brought up that provoked this response include Tiananmen Square, Winnie the Pooh, Taiwan’s controversial independence, and Maoism. Furthermore, when asked about what morals China was built on, it gave a list which initially started out with Confucianism but then briefly mentioned socialism before promptly censoring itself.

But perhaps most importantly is the subject of any possible data concerns and censorship. Firstly, the app has no major data or security risks publicly known at this time. However, this has not stopped United States Legislation from promptly questioning it. Currently, the state of New York has already banned the app, calling to attention the Chinese laws regarding data where any Chinese company must cooperate with the CCP government if asked for any user data. This reaction is quite similar to the early success of another Chinese-owned app, TikTok in the United States. 

Furthermore, Congress also already introduced a bipartisan bill to ban the DeepSeek app off of government devices. Republican senator Josh Hawley has also introduced a bill late last January that would ban Americans from downloading and using any Chinese AI models in general, which raises concerns about various freedoms. Hawley has been quoted saying “every dollar and gig of data that flows into Chinese AI are dollars and data that will ultimately be used against the United States.”  The State of Virginia has also taken action and has banned the app off of state devices, coming after the governor Glen Youngkin raised concerns about alleged hidden code that could serve to secretly share data. Despite all of these concerns, it may also be worth noting that the model, unlike Chat GPT, is open-source, which means that many factors of code and inner workings are publicly viewable and even modifiable. 

For now, the future for the AI model’s presence seems quite uncertain and the timeline of events seems to be following TikTok’s controversial restrictions. Currently, it would be smart to stay informed and engaged in any state and national restrictions of any kind