DeepSeek AI ban: What's going on and why it matters

Chinese artificial intelligence (AI) startup DeepSeek that has rocked the global AI landscape by recently launching R1 LLM model, is facing crackdowns from several governments, including India, the US, Australia, Taiwan, and South Korea, due to concerns over its security and data practices, particularly on government devices. While most firms are still trying to understand the ramifications of new open-source models like DeepSeek R1, many have also refrained from using the model, citing privacy and security concerns.

Which countries and agencies have banned DeepSeek?

Over the last few days, DeepSeek AI has been banned in multiple countries and agencies due to data privacy and security concerns linked to potential access by the Chinese government. Italy, Taiwan, Australia, South Korea, India, and the US have all restricted or prohibited its use.

Italy was among the first to ban DeepSeek AI, citing issues with user data handling and compliance with EU data protection laws. The Italian Data Protection Authority removed the app from stores after investigating its data collection practices.

Taiwan has prohibited DeepSeek's use in all public sector organisations, including schools and critical infrastructure, due to worries about cross-border data transmission and information leaks. Australia has barred its employees from using DeepSeek on government devices, with Home Affairs Minister Tony Burke stating that a national intelligence assessment deemed it an unacceptable security risk.

South Korea's defense ministry has blocked DeepSeek from its military computers, following a request for clarification on the app's user information management from the personal information protection commission.

In the United States, the Navy and other federal agencies restricted DeepSeek's use, and Texas was the first state to ban the app. Access to DeepSeek technologies has been blocked in the Pentagon, which is the headquarters of the US Department of Defense (DoD) and space agency NASA, who have instructed employees to avoid DeepSeek, and numerous companies have asked cybersecurity firms to block it.

India's Ministry of Finance has prohibited DeepSeek for its employees, extending a ban on AI tools like DeepSeek and ChatGPT on government devices.

Why have some countries banned DeepSeek?

Countries blocking DeepSeek cite security risks associated with the Chinese application and insufficient information on how user data will be stored or used. DeepSeek’s privacy policy indicates it collects personal information, chat history, and technical data, sharing it with service providers and retaining it as long as necessary.

Feroot Security, a cybersecurity firm, recently revealed hidden code in DeepSeek that allows direct data transmission to China Mobile servers, a state-owned company controlled by the Chinese government. Users registering on DeepSeek may unknowingly expose their data to Chinese state systems.

Experts verified Feroot's findings, which indicated that DeepSeek could send user data to CMPassport.com, an online registry for China Mobile.

What happens when a ban is enacted on DeepSeek’s models?

While the investigation continues, experts believe a ban could impact users and enterprises by restricting access, creating compliance risks, and causing data continuity issues.

Adrianus Warmenhoven, a cybersecurity expert at NordVPN, noted that DeepSeek's privacy policy states user data, including conversations and responses, is stored on servers in China. This raises concerns due to data collection practices that involve both user-shared information and data from external sources, highlighting risks associated with differing privacy and security standards in that jurisdiction.

According to a security specialist of a mid-sized hospital in India, the ban might require urgent migration away from cloud-based deployments on platforms like Azure, AWS, and Nvidia, potentially leaving DeepSeek outdated and vulnerable.

Why banning DeepSeek is harder than it looks

Banning access to the DeepSeek website and mobile applications may sound simple, but there are challenges in the process believes experts. That’s because the DeepSeek LLM itself is open-source, blocking access to it is less straightforward, as it can be run locally on a device, it can be run through a cloud environment, and installed via various apps or frameworks like Ollama. It requires some extra leg-work for users to set-up, but it can be done, believes Satnam Narang, Senior Staff Research Engineer at Tenable.

That said, the local LLMs don’t require access to external servers, minimising the risk of leaking sensitive data. However, as Narang noted banning DeepSeek’s website and apps on government devices is one barrier, because they are still accessible on personal devices, which could be used to share sensitive information.

Another concerning aspect to DeepSeek is not just the exposure of sensitive information, but rather the model’s lack of safety features, as it could be used for harmful or nefarious purposes, believe experts.

What’s the future of DeepSeek?

Despite government bans and regulatory hassles, DeepSeek – an answer to existing LLMs such as ChatGPT and Gemini, has shown how small, focused teams with limited resources can drive tech breakthroughs.

While DeepSeek has taken steps to address the growing number of bans and restrictions by updating its privacy policies to provide greater transparency regarding data collection, storage, and usage, many governments and organisations remain cautious, emphasising the need for rigorous assessments to ensure that DeepSeek’s operations do not compromise national security or user privacy.

These actions reflect a growing global apprehension regarding data handling practices and their potential implications of AI models for national security and user privacy. Organisations and governments are increasingly cautious about integrating foreign-developed AI technologies into their systems without thorough assessments of associated risks.

Nonetheless, with the AI landscape being dominated by "Big Tech" companies in the US, with the emergence of DeepSeek’s disruptive AI model, it is likely that smaller companies globally including India will have a growing role to play in creating AI tools that have the potential to make our lives easier.