DeepSeek, a rapidly rising AI startup in China, has reportedly asked key employees to surrender their passports to prevent the risk of sensitive information being leaked.

Earlier this year, DeepSeek made headlines when it emerged as a potential competitor to leading Western AI companies like OpenAI and Google DeepMind.

The startup has released a high-performance, cost-effective AI model with standout features like chatbots, content generation, and many other AI tools.

The overwhelming success of this model helped DeepSeek quickly take the number one position on major app stores, making the company a notable name on the global AI map.

In China, this achievement also helped DeepSeek be considered a “national treasure”.

deepseek.jpg
DeepSeek is considered a "national treasure" in China. Photo: Bloomberg

However, according to anonymous sources from The Information, it is this rapid growth that has led to tighter controls.

Key DeepSeek employees were required to surrender their passports to prevent the leak of “internal information that could constitute trade secrets or even state secrets.”

The move reflects the Chinese government's keen interest in protecting strategic technology assets amid intensifying global AI competition.

International concerns about data security

As DeepSeek's application grew in popularity around the world, concerns about data security and Chinese government influence began to emerge.

Experts and policymakers question whether user data from this AI model could be accessed by Chinese authorities.

The US, European Union and India have all expressed concerns about DeepSeek's links to China.

Specifically, Washington has imposed restrictions on Chinese AI companies, citing national security risks, while the European Union is considering regulatory measures for AI services that handle sensitive data.

India also recently announced a review of policies related to foreign AI applications operating in the country. These developments show the growing wariness of the international community towards Chinese technology companies, especially in the field of AI.

Domestic competitive pressure

DeepSeek is not only facing international pressure, but also fierce competition at home. Last week, Chinese tech giant Baidu announced a new inference AI model called X1, which it claims has comparable performance to DeepSeek’s model but at a lower cost.

Along with the Ernie 4.5 platform model, Baidu has integrated these technologies into the Ernie Bot chatbot and made it available to users for free, two weeks earlier than originally planned.

Previously, users had to pay a monthly subscription fee to access Baidu's latest AI models.

In an announcement on WeChat, Baidu said it plans to open source its Ernie AI models from June 30, a move aimed at increasing competitiveness.

As one of the pioneers in China to launch a generative AI platform from 2023, Baidu is now under great pressure from domestic rivals such as ByteDance (owner of TikTok) and Moonshot AI, which have attracted a large number of users with their chatbots.

The emergence of DeepSeek in January, with an AI model that performs on par with ChatGPT but costs much less, has changed the competitive landscape in China.

Analysts say DeepSeek's low-cost, open-source strategy is forcing leading AI companies like Baidu to adjust their strategies, including cutting prices and accelerating product releases to maintain market share.

DeepSeek Founder Liang Wenfeng: China's New Hero When Liang Wenfeng returned to his hometown for the Lunar New Year, a crowd gathered around his house, including many parents with young children, hoping that they would follow the example of DeepSeek's founder.