A recent analysis by Wired US revealed that DeepSeek, an innovative artificial intelligence service, appears to be transmitting data to both Chinese tech giant Baidu, presumably for web analytics purposes, and to Volces, a Chinese internet infrastructure company.
Article Key Points
Collecting data from multiple sources
DeepSeek also reserves the right to collect data from third-party sources. For example, if a user creates an account using Google or Apple sign-on, the company will have access to some information provided by these giants. Additionally, advertisers can share data with DeepSeek, which may include “mobile advertising identifiers, encrypted email addresses and phone numbers, and cookie identifiers.”
DeepSeek's Use of Information
The startup says it will use the information it collects for a variety of purposes, including “maintaining the service, enforcing our terms and conditions, and making improvements.” However, the document also mentions that DeepSeek may use user prompts to develop new models. According to the terms of use, DeepSeek will “analyze, improve, and develop the service, including by tracking interactions and device usage, analyzing how people use it, and training and improving our technology.”
DeepSeek also notes that the information collected will be used to “comply with its legal obligations,” a broad clause that many companies include in their policies. This data may be accessed by the “corporate group” and shared with law enforcement, public authorities, and other entities when required.
Impacts of Chinese Privacy Laws
Over the past decade, China has enacted several cybersecurity and privacy laws, allowing state authorities to demand information from tech companies. A 2017 law, for example, requires organizations and citizens to “cooperate with national intelligence efforts.” These regulations, along with U.S.-China trade tensions, have raised concerns about the security of apps like TikTok. Critics of the platform say it could collect massive amounts of data and send it to China, potentially influencing users with Chinese propaganda.
Expert Olejnik points out that, compared to users of a social network like TikTok, those who interact with a generative AI are more actively involved. This means that the influence they are subject to could be greater, “especially considering that the internal workings of the model are largely unknown.”
Lawmakers in several countries could take similar measures to those against TikTok against AI companies. Olejnik says: “We cannot rule out that 2025 will bring an expansion of direct actions against AI companies… Data collection could once again be cited as the reason.” In Italy, there are already concerns about DeepSeek’s possible non-compliance with the GDPR, the European Union’s privacy regulation, while rumors are spreading about an imminent intervention by the privacy regulator, similar to the one already taken against OpenAI.
Article published on 28 January 2025 - 21:46
The article is interesting, but I wonder if privacy laws in China are really as restrictive as they say. It is important that user data is protected, otherwise there is a risk of abuse by companies.