Opinion: Seeking Clarity on DeepSeek for Higher Education

In January 2025, the Chinese AI company DeepSeek roiled U.S. financial markets and the tech industry by announcing it had developed a large-scale AI language model at a fraction of the current development costs. On Jan. 27, Nasdaq dropped 3.4 percent at the market open, with Nvidia declining by 17 percent, losing a staggering $600 billion in market capitalization. DeepSeek, an AI development company in Hangzhou, China, founded by Liang Wenfeng, is a direct challenge to the established AI industry leader ChatGPT. DeepSeek’s new open-source AI model called R1 has been described as having a “new level of reasoning.” In January, TechCrunch reported that DeepSeek's "reasoning" model could beat OpenAI’s on certain benchmarks.

The introduction of DeepSeek has the potential to positively influence corporate and educational sectors, but at the same time poses some dangerous possibilities.

COST AND USE DISRUPTION

One of the key features of DeepSeek is the claim that their R1 model was created for less than $6 million, significantly less than the billions of dollars invested in AI by U.S. tech companies. Another is that the company was able to create the product without high-performance AI accelerators and GPU chips from the U.S., which have been restricted in China.

The spike in popularity of DeepSeek has been dramatic, as it has taken over ChatGPT’s spot as the most-downloaded free app in Apple’s App Store, according to CNBC. Some articles in the press have referred to the introduction of DeepSeek as the “Sputnik Moment,” referencing the Soviet Union’s launch of Sputnik 1, the first Earth-orbiting satellite in 1957 that escalated the space race. The questions for business and education sectors are:

1. If DeepSeek was developed at such a low cost, can it also accurately and effectively perform as well or better than other, proven AI products?
2. Is it cyber secure?
   

POSITIVE FEATURES

Beyond the potential financial advantages of DeepSeek, another important aspect is that it is based on open-source code. As explained in a recent piece published by the World Economic Forum, “[DeepSeek] is available for anyone to download, copy and build upon. Its code and comprehensive technical explanations are freely shared, enabling global developers and organizations to access, modify and implement.” As an open-source tool, DeepSeek is freely available to everyone, including researchers and developers.

However, users should still be cautious, since AI tools can quickly change as they are improved and developed. Updates to DeepSeek are all but certain. A recent piece from the tech journalism outlet VentureBeat put it this way: “ChatGPT’s vision and image generation capabilities are still hugely important and valuable in workplace and personal settings — DeepSeek-R1 doesn’t have any yet.”

NEGATIVE ASPECTS

There are some glaring negative aspects of DeepSeek which should cause some end users to be cautious. First, as a Chinese product, DeepSeek is subject to country-specific censorship and typically mirrors Chinese Communist Party lines. According to The Guardian, when DeepSeek is asked about a specific topic on controversial Chinese subjects such as Tiananmen Square, it responds by replying, “Sorry, that's beyond my current scope. Let's talk about something else.”

CYBER RISKS

One of the most important concerns around DeepSeek is the cyber risk. Since this product is housed in China, there are serious concerns about data privacy, where the data is stored and how it is utilized. Almost as soon as DeepSeek was available for download, it was targeted by distributed denial-of-service (DDoS) cyber attacks on Jan. 27, 2025. According to a news report from TechTarget, “The top three sources of attack infrastructure were the U.S. (20 percent), the U.K. (17 percent) and Australia (9 percent).” The global concern around DeepSeek’s potential ability to harvest and steal data should be a caveat for any potential user.

IMPLICATIONS FOR GOVERNMENT AND EDUCATION

Certainly, U.S. governmental use of DeepSeek is suspect, given the risks involved. New York Gov. Kathy Hochul already has announced a statewide ban. According to a news release from her office, "Serious concerns have been raised concerning DeepSeek AI’s connection to foreign government surveillance and censorship, including how DeepSeek can be used to harvest user data and steal technology secrets.” The cybersecurity news website GovInfoSecurity echoed these concerns, pointing out weak encryption, data transfers to China, and the app's ability to send unencrypted data to servers controlled by ByteDance, the company that owns TikTok.

As a free open-source tool, DeepSeek could be attractive for users in the educational sector. Teachers and students could use it for research, curriculum planning and development, and for supporting many academic activities. Having the ability to customize the application for myriad potential uses is attractive. One resource, Kangaroos.ai, has compiled a list of ways DeepSeek could be utilized. Still, for all the potential advantages of DeepSeek, an equally alarming number of negatives stand in the way of its worldwide adoption. Ensuring an AI tool is factual, non-biased, and uncensored will be important for higher education. While we try to seek clarity around DeepSeek, we must also look at all AI tools to ensure they are ethically and safely used, including ChatGPT, Microsoft’s Copilot and others.
 

WHAT'S NEXT

We know AI tools will continue to evolve, and new products will be developed, distributed and analyzed. DeepSeek may have ushered in a new era and made more space for smaller AI companies to push the industry even further as they introduce their own new innovations globally. At the recent AI Action Summit in Paris, world leaders and tech CEOs discussed five important themes in AI, including public interest, jobs, investment, ethics and regulation. A variety of countries announced their own global plans for AI. In fact, President Emmanuel Macron of France encouraged people to download their new AI chatbot, Le Chat, which is purported to be 13 times faster than ChatGPT. This signals that there will be increased global competition to make AI tools faster, cheaper and more immersive. Business, government and education sectors will need to sort out what AI can and cannot do, with an eye on its appropriate and ethical use. Each sector will need to seek their own clarity in this complex AI environment.