San Anselmo, California – January 27: In this photographic illustration, the Deepseek app appears in … [+]
What Deepseek-R1’s privacy policy reveals to its systems it deserves great attention and utmost care in its use. It is not about the technological ability or exceeding 01 of Openai or others in the standards related to mathematics, coding or general knowledge – goals already discussed.
It is about her ability to display artificial integrity on intelligence.
First, the lack of Deepseek’s internal – ‘inner’ internal mechanisms makes it a bigger use model than user empowerment.
Deepseek’s privacy policy describes the types of data it collects, but fails to clarify how these data are processed inside. Users’ inputs such as conversation history and loaded files have been collected to “train and improve services”, however no mention is mentioned by anonymity or protective measures for sensitive data.
There is no clear documentation if user data is used directly to update it. Terms such as “Hashed emails” and “mobile identifiers” darkening significant transparency, leaving the user unsafe about the implications of their data collected.
In general, Deepseek collects extensive data (eg, keys, device ID) but does not justify how these grain details are needed to provide its service. However, it takes care to show that it is keeping user data “as long as it is necessary” but without specific retention periods or guarantees user data exposure to prolonged weaknesses, including misuse, violations or access to unauthorized access .
Its reliance on tracking mechanisms (such as cookies) demonstrates a basic trading: users can “disable cookies”, but the policy warns that it limits functionality, forcing users in sharing data for basic service use. Furthermore, by linking essential functions such as inputs or continuity of the account with data collection practices, Deepseek blurs the line between informed consent and forced compliance.
Included, its policy does not mention any mechanism to prevent prejudice in the way the system processes user inputs or generates responses, while there is no mention of explanation in the way it is generated, leaving the user in the dark regarding logic after decisions or recommendations.
And lastly, relying on internal reviews of user inputs to implement “Terms of Service”, Deepseek sets an ethical behavior on users, not the system itself.
Second, the promises of Deepseek’s innovation should not justify his mistakes on critical external issues – outer ‘external’ dangerous structures.
Deepseek stores personal information on servers located in the People’s Republic of China, and its privacy policy accepts cross -border data transfers.
While mentioning legal compliance, there is no clear mention of compliance with the main global frameworks of intimacy such as GDPR (Europe) or CCPA (California), raising concerns for the legal treatment of user data from strict data protection jurisdictions .
Given the regulatory environment in China, where data localization and government access are important concerns, storing personal data sensitive to Chinese servers presents potential geopolitical weaknesses as users from regions with strict data protection laws can Not to undergo less protective data regimes, undermining their privacy rights
Deepseek openly acknowledges user data sharing with advertising and analytics partners to earn its platform before enabling them to target users based on granular data, including out -of -platform activities.
And typical for such a model, there is little (if any) transparency about how users’ compensation – or even informed. Not to mention that the data collected can be used to perpetuate existing inequalities, such as the target of vulnerable populations with manipulative advertising. Indeed, while algorithms form what users see and consume, they indirectly affect social behaviors, values and tendencies, often in ways that prioritize profit on well -being.
Privacy policy also allows Deepseek to divide user data during corporate transactions, such as merger, purchases or sales, leaving the user data touched for further use, for which they sign an empty check .
And it is worth noting the lack of independent audits or external validity, which means that users must rely on the self-regulation of Deepseek-a dangerous proposal for any system.
Third, in the failure to address the weaknesses in the relationship – ‘Inter’ – Deepseek risks turning from a mediator to a predator.
Deepseek’s policy positions user participation as a contingent in important data exchange.
For example, while users may disable cookies, they are warned that this will result in reduced functionality, forcing them effectively in sharing data for a “smooth” experience.
Although users can delete their data, policy offers little clarity on the consequences of using long -term service, creating an imbalance in the relationship between the platform and its users.
Moreover, his user’s input treatment, such as the story of the conversation and the charged files, raises considerable concerns about how the platform mediates human relationships. Indeed, the data provided by the user is treated as a source for platform benefit (eg, model training), without clear Opt-out opportunities for individuals who do not want their data to be used this way .
While Deepseek users states can exercise rights such as data deletion or access, the process is buried under layers of verification.
Also, the announcement of the intimacy of the platform does not provide any guarantees that the answers or results of it are rooted in the principles led by integrity, leaving the user unsafe for the reliability of the interactions.
Equally disturbing, treating Deepseek dependent relationships, such as minors or emotionally tangible users, underlines critical supervisions in his mediation mechanisms.
While politics acknowledges parents’ consent to users under 18, it lacks strong protective measures to prevent data misuse or use of new users. It is not mentioned how Deepseek systems detect or treat users anxious, such as those who discuss mental health or other sensitive issues, creating a risk of emotional damage.
Finally, regular privacy policy updates are mentioned, but there is no clear process for users to trace changes that may significantly affect their intimacy.
Refining what we ask of it – artificial integrity on intelligence – is the guarantee of the performance of it by addressing the service of what matters most: Humanity. Without this, economic value comes at the expense of social welfare and therefore, individual life.
It needs performance that does not come to the expense of excess energy, water and land resources, nor lead to economic concentration in the hands of some.
It should also be integrated with integrity, not just from an external perspective, but mainly in its essential functioning. Without this, artificially created intelligence can go to harmful territory at a social level, beyond what any developer could manage with a return.
In the first, let’s hope that the promise of models such as the Deepseek-R1 opens the basic ways; While most importantly, in the latter, ensuring that innovation empowers people with cars, not vice versa – artistic integrity over intelligence.