Governments in the U.S., Italy, and now Australia have started blocking access to DeepSeek, a Chinese-developed large language model (LLM), on government devices. The concern? Data security. Officials fear that sensitive information could be exposed to the People’s Republic of China (PRC). While banning DeepSeek’s website and mobile apps is relatively simple, the real challenge comes from the fact that the model is open-source.
Because DeepSeek is open-source, it can be run locally on a device, through cloud environments, or integrated into other apps. This means that even if government agencies block the website and apps, it can still be accessed in ways that are harder to control. DeepSeek doesn’t require external servers, which lowers the risk of data leaks. But once it's installed, it can be used offline, potentially bypassing restrictions.
Satnam Narang, Senior Staff Research Engineer at Tenable, points out another problem: the lack of safety features in DeepSeek. It’s not just about data exposure; the model could also be misused for harmful purposes. This makes it even more concerning for governments trying to protect sensitive information.
In short, while banning access to DeepSeek's apps and websites is an important step, the open-source nature of the model makes it much harder to fully block. Governments will have to come up with more comprehensive measures to address these emerging risks.