DeepSeek, the Chinese AI start-up, has generated significant attention since its launch of the R1 chatbot in January. Despite a claimed smaller budget than industry giants, the company asserts that its AI can rival those from the US at a fraction of the cost. As its influence grows, concerns over data security and international competition are intensifying, however, leading to a wave of regulatory action in several nations.

Taiwan’s decision to ban government agencies and critical infrastructure from using DeepSeek, citing national security risks, is just one example of the growing pushback against the company. On February 1, Taiwan’s Ministry of Digital Affairs cited concerns about sensitive data leakage, reflecting broader fears over Chinese technology. This decision follows Taiwan’s ongoing wariness about Chinese tech, especially since 2019, when it began restricting foreign technology that could compromise national security.

Taiwan is not alone in voicing concerns. South Korea’s government temporarily banned DeepSeek from its ministries, including those dealing with foreign affairs and national defence. The country’s information watchdog, the Personal Information Protection Commission, has launched an investigation into how DeepSeek handles user data. Similarly, Italy’s data protection authority, Garante, has raised alarms about DeepSeek’s lack of transparency in its data practices, even going as far as restricting the processing of Italian users’ data.

Countries banning DeepSeek's AI program

Multiple countries have either enacted or are planning to enact bans on DeepSeek’s AI technology, primarily due to security and data privacy concerns.

United States
In the US, lawmakers are reportedly preparing legislation to block DeepSeek from government devices. NASA followed suit on January 31, prohibiting its use on agency devices. A week earlier, the US Navy warned members against using the platform due to “security and ethical concerns.”

Italy
On January 30, Italy’s Garante ordered a temporary ban on the processing of Italian user data by DeepSeek, raising concerns about how the company stores and handles personal information. Italy’s scrutiny follows broader European concerns about Chinese tech companies.

South Korea
South Korea banned DeepSeek from government devices on January 31, citing security concerns. The Ministry of Trade, Industry, and Energy followed up with a broader warning to government agencies about using AI programs like DeepSeek, emphasising the need for caution over data management.

Australia
Australia’s government also blocked DeepSeek on all government devices on February 3. The Department of Home Affairs cited security risks, directing all government entities to remove any DeepSeek products from their systems.

Taiwan
Taiwan’s ban on DeepSeek, issued on February 1, further highlights security concerns. Government agencies and critical infrastructure were prohibited from using the software due to fears of sensitive data leakage.

Why place bans on DeepSeek?

The central concern in these countries is security - particularly regarding DeepSeek’s handling of user data. The company collects personal data such as emails, phone numbers, and dates of birth, as well as chat history and technical information, including IP addresses and device details. Many governments fear this data could be exploited or used for surveillance purposes, especially given DeepSeek’s Chinese origin and the links of its founder to the Chinese government.

Founded in 2023, DeepSeek has quickly positioned itself as a formidable competitor in the AI space. Despite being produced - according to Chinese media at least - on a fraction of the budget of its American counterparts, the company has claimed to develop a similarly powerful AI model for far less. While American companies like OpenAI spend billions on AI research, DeepSeek has developed its technology with a budget of only around $190mn if reports from Beijing are accurate.

Furthermore, DeepSeek’s approach to AI technology differs in key ways, including the use of a technique called Mixture of Experts (MoE). This model allows DeepSeek’s AI to activate specific sub-models for different tasks, making it more efficient and less costly to run than traditional AI systems. This model helps reduce computational expenses and energy usage, offering the company an edge in terms of cost-efficiency.

Another major differentiator is DeepSeek’s open-source approach. Unlike competitors that restrict access to their AI systems and charge high fees, DeepSeek makes its technology available to developers at a much lower cost. API calls to DeepSeek’s models can cost as little as $1-2, compared to OpenAI’s price of $15-60. This open-access model has thus made DeepSeek particularly attractive to smaller companies and startups, potentially helping to democratise access to advanced AI.

The open-source model also benefits DeepSeek by allowing global collaboration in the AI community. External developers can contribute to its development, speeding up innovation and improving the technology without the need for extensive in-house research teams.

However, despite its low cost and innovative approach, DeepSeek faces mounting resistance. Many governments are concerned that its AI could serve as a tool for surveillance or data harvesting, particularly as China tightens its grip on tech regulation and data policies.

These concerns have prompted calls for greater scrutiny of Chinese-made technologies, particularly as geopolitical tensions between the US and China intensify. In East Asia too, the DeepSeek response to questions on Taiwan as a self-governing nation, and the Chinese military massacre of students in Tiananmen Square in 1989 has already led to posts on X and BlueSky showing DeepSeek seemingly confused on how to respond and asking the original poster to change the topic.

Despite these hurdles, DeepSeek’s ability to provide accessible, affordable, and efficient AI is undeniable. If DeepSeek continues to address data privacy concerns and scales its operations effectively, it could become a dominant player in the global AI market.

Whether DeepSeek can overcome regulatory challenges put in its path around the world, and continue to disrupt the market remains to be seen, but it’s clear that its presence is already reshaping the AI landscape for better—or worse. It just depends who you ask.