The rise of DeepSeek, the Chinese AI chatbot, has sent ripples throughout the global tech landscape, disrupting markets and raising concerns about national security. Its R1 program, reportedly built using less advanced Nvidia transistors, has demonstrated impressive capabilities, shaking up the dominance of established players and triggering a cascade of bans by governments worldwide.
Indeed, the rapid rise of artificial intelligence presents both exciting possibilities and challenges. But is this a justified security measure or a symptom of escalating tech rivalry?
DeepSeek’s emergence has been nothing short of disruptive. Its cost-effectiveness and performance have challenged the status quo, impacting US tech equities and forcing a reassessment of the competitive landscape. However, this technological leap has been met with suspicion and swift action by several nations.
Italy was the first to raise a red flag, launching an investigation into DeepSeek and barring it from accessing Italian user data. This move echoed Italy’s earlier temporary ban on ChatGPT, highlighting growing anxieties about data privacy in the age of generative AI.
Taiwan, citing national security risks, prohibited DeepSeek’s use by public sector employees and at critical infrastructure facilities. Australia followed suit shortly after, further amplifying concerns about the Chinese chatbot.
The bans haven’t stopped there. South Korea, a key US ally, has restricted DeepSeek’s use within its police force and numerous government ministries, including defense and unification, the latter dealing with sensitive relations with North Korea. In the US, Congressman Darin LaHood introduced the “No DeepSeek on Government Devices Act,” labeling the “Chinese Communist Party-affiliated company” as an “alarming” national security threat.
DeepSeek and data security concerns
The core of these concerns revolves around data security and potential government access.
DeepSeek’s terms and conditions, like those of OpenAI’s ChatGPT, include provisions for sharing personal information with third parties. However, the crucial difference lies in the legal obligations imposed on companies in China.
As Youm Heung-youl, a data security professor at Soonchunhyang University, explains, “In China, when the government requests access, companies are legally obligated to provide user data,” a stark contrast to the often-contentious relationship between US companies and government data requests.
This fundamental difference in legal frameworks fuels suspicions about DeepSeek’s data handling practices. Further adding to the unease is DeepSeek’s collection of “keystroke patterns or rhythms,” a practice outlined in its privacy statement.
While Beijing insists that it “will never require enterprises or individuals to illegally collect or store data,” these assurances have done little to quell anxieties.
Guo Jiakun, a Chinese foreign ministry spokesman, dismissed the restrictions as “the politicization of economic, trade, and technological issues,” but the concerns persist. For some, the bans appear to be a knee-jerk reaction to US influence.
Vladimir Tikhonov, a professor of Korea studies at the University of Oslo, suggests that South Korea’s restrictions reflect “both genuine concerns and their knee-jerk reflective response to follow the US line.” He points to the close security alliance between the two countries, noting that South Korea “can’t operate alone in a situation [of] large-scale warfare” and is expected to align with US cybersecurity guidelines.
However, Tikhonov also acknowledges the inherent political realities of the tech world, stating that “no big tech company is politically neutral.” He points to Google’s data collection practices, suggesting that while US companies may resist government requests, data sharing does occur. He implies that collaboration between Chinese internet companies and the government is likely “more extensive.”
Isabel Hou, a Taiwanese AI expert, highlights another dimension of the issue. She claims that DeepSeek “has a policy of aligning with the core values of socialism” and applies these rules even when providing services overseas. This raises questions about censorship and access to information, particularly on sensitive topics like Taiwanese sovereignty or the Tiananmen Square massacre.
The rapid rise of DeepSeek
The rapid rise of DeepSeek has also surprised many. Park Seung-chan, a professor of Chinese studies at Yongin University, notes that “DeepSeek was launched in May of 2023, and something like this can’t just emerge overnight.” Experts point to China’s substantial investments in research and development as a key factor in DeepSeek’s rapid advancement. China’s R&D spending has grown exponentially in recent years, placing it second only to the US globally.
DeepSeek’s reported use of less-advanced H800 chips, allowed for export to China until 2023, has further complicated the situation.
Park Ki-soon, a professor of Chinese economics at Sungkyunkwan University, suggests that this demonstrates that “even without cutting-edge semiconductors, similar outcomes could be achieved with general semiconductors, as long as the software is good.” This underscores the importance of software development and the need for governments to invest in this crucial area.
The DeepSeek saga raises fundamental questions about the balance between national security and technological innovation. While concerns about data security and government access are valid, the bans also raise the specter of protectionism and a potential chilling effect on AI research.
As the AI landscape continues to evolve, navigating these complex issues will be crucial for governments and businesses alike. The future of DeepSeek, and the broader implications for global tech competition, remain to be seen.
