澳洲表示，使用 DeepSeek 時要小心——那麼它使用安全嗎？

DeepSeek is a new and popular app that has recently taken the app stores by storm, especially in Australia and the United States. It has reached the top position in the app store rankings, which is quite an achievement. However, this rise to fame has not come without its share of concerns. Ed Husic, the science minister of Australia, has become the first official from a Western government to express worries about the privacy implications of using DeepSeek. This Chinese chatbot has stirred up quite a bit of controversy in the tech industry and financial markets. There have been ongoing allegations regarding Chinese technology companies, such as Huawei and TikTok, being closely linked to the Chinese government. This has led to fears that personal data collected by these companies could be used for intelligence purposes. While former President Donald Trump mentioned DeepSeek in relation to the US, he did not categorize it as a national security threat. Instead, he suggested that it could potentially lower costs for consumers. However, Husic emphasized the need for caution, stating that there are still many unanswered questions regarding data and privacy management. He advised the public to be careful when using the app, as these issues require careful consideration. DeepSeek has not yet responded to inquiries from the BBC regarding these privacy concerns, but users in both the UK and US seem to be using the app without hesitation. According to market analysts at Sensor Tower, DeepSeek has achieved an impressive three million downloads since its launch, with a staggering 80% of those downloads occurring in just the past week. This rapid growth indicates that DeepSeek is being downloaded at a rate three times faster than its competitors, such as Perplexity.

One of the key concerns surrounding DeepSeek is the amount of personal information it collects from its users. According to the app's privacy policy, it gathers a wide range of data, including email addresses, phone numbers, and dates of birth when users create their accounts. Additionally, it collects any user input, such as text and audio, as well as chat histories. The app also gathers what it refers to as 'technical information,' which includes details about the user's phone model, operating system, IP address, and even keystroke patterns. DeepSeek claims that it uses this information to enhance the app's safety, security, and stability. However, it also shares this data with third parties, including service providers and advertising partners, and retains it for as long as necessary. Lauren Hendry Parsons, a digital privacy advocate at ExpressVPN, has raised genuine concerns about the technological potential of DeepSeek, particularly regarding its privacy policy. She pointed out a specific clause that states the app can use data to help match users and their actions outside of the service, which she believes should raise alarm bells for anyone concerned about their privacy. While DeepSeek does collect a significant amount of data, experts note that its privacy practices are not entirely different from those of other popular services like ChatGPT, Gemini, or even social media platforms.

So, is DeepSeek safe to use? Emily Taylor, the chief executive of Oxford Information Labs, explained that for any AI model that is openly available, including DeepSeek, the prompts or questions asked by users become accessible to the creators of that model, along with the corresponding answers. This means that individuals working on confidential or national security matters need to be aware of the potential risks involved. Dr. Richard Whittle from the University of Salford expressed his concerns about data and privacy with the app, but he also acknowledged that there are similar issues with models used in the US. He advised consumers to remain cautious, especially in the face of the hype and fear of missing out on a new and highly popular app. The UK data regulator, the Information Commissioner's Office, has urged the public to be mindful of how their information is used to train AI models. When asked if they shared the Australian government's concerns, they stated that generative AI developers and deployers must ensure that individuals have meaningful, concise, and easily accessible information about the use of their personal data. They also emphasized the importance of having clear and effective processes for enabling individuals to exercise their information rights. The Information Commissioner's Office will continue to engage with stakeholders to promote effective transparency measures and will not hesitate to take action if regulatory expectations are ignored.